Add openenv-core dependency and server entry point

.build-trigger

@@ -1 +1 @@
-Final Release Sync (Definitive UI): 2026-04-07 23:38:07
+Final Release Sync (Definitive UI): 2026-04-07 23:38:07

.dockerignore

@@ -1,10 +1,10 @@
-.git/
-.env
-backend/venv/
-backend/__pycache__/
-frontend/node_modules/
-frontend/dist/
-.pytest_cache/
-.coverage
-brain/
-.gemini/
+.git/
+.env
+backend/venv/
+backend/__pycache__/
+frontend/node_modules/
+frontend/dist/
+.pytest_cache/
+.coverage
+brain/
+.gemini/

.gitignore

@@ -1,37 +1,37 @@
-# Python
-__pycache__/
-*.py[cod]
-*$py.class
-*.so
-.Python
-backend/venv/
-.pytest_cache/
-.coverage
-.cache
-backend/scenarios/.cache
-
-# Node
-node_modules/
-.npm/
-
-# Env & Secrets
-.env
-.env.*
-!.env.example
-# default.env is needed for HF Spaces
-
-# OS
-.DS_Store
-Thumbs.db
-
-# VS Code / IDE
-.vscode/
-.idea/
-*.swp
-*.swo
-
-# Project specific
-backend/logs/
-*.log
-.gemini/
-brain/
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+backend/venv/
+.pytest_cache/
+.coverage
+.cache
+backend/scenarios/.cache
+
+# Node
+node_modules/
+.npm/
+
+# Env & Secrets
+.env
+.env.*
+!.env.example
+# default.env is needed for HF Spaces
+
+# OS
+.DS_Store
+Thumbs.db
+
+# VS Code / IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+
+# Project specific
+backend/logs/
+*.log
+.gemini/
+brain/

README.md

@@ -1,292 +1,292 @@
----
-title: NEXON-AI
-emoji: 🛡️
-colorFrom: blue
-colorTo: indigo
-sdk: docker
-app_port: 7860
-pinned: false
----
-
-<!-- LAST_SYNC_VERIFICATION: 2026-04-08 00:07:00 -->
-
-# NEXUS-AI 🌐🛡️
-### Autonomous Incident Investigation Dashboard
-
-<div align="center">
-
-![Python](https://img.shields.io/badge/Python-3.10+-3776AB?style=for-the-badge&logo=python&logoColor=white)
-![FastAPI](https://img.shields.io/badge/FastAPI-0.100+-009688?style=for-the-badge&logo=fastapi&logoColor=white)
-![React](https://img.shields.io/badge/React-18.x-61DAFB?style=for-the-badge&logo=react&logoColor=black)
-![Tailwind](https://img.shields.io/badge/Tailwind_CSS-3.x-38B2AC?style=for-the-badge&logo=tailwind-css&logoColor=white)
-![Ollama](https://img.shields.io/badge/Ollama-Local_LLM-000000?style=for-the-badge&logo=ollama)
-
-**Status:** Active Simulation Pipeline  
-**Architecture:** Real-time WebSockets + Multi-Agent Consensus
-
-</div>
-
----
-
-## 📖 What is NEXUS-AI?
-
-NEXUS is a next-generation, autonomous dual-agent environment designed to investigate and validate software incidents in real-time. Using a combination of an **Investigator** and a **Validator** agent, NEXUS autonomously forms hypotheses, executes systems tools, evaluates system behavior, and reaches strict consensus on root causes.
-
-Traditional manual debugging requires extensive context-switching and tool fatigue. NEXUS solves this through:
-1. **Dual-Agent Autonomy**: Two specialized models communicating word-by-word via WebSockets.
-2. **Dynamic Tool Execution**: Fully integrated system terminals allowing agents to run sandboxed validation scripts.
-3. **Semantic Reward Engine**: Evaluates conversational drift mathematically (using native GPU embeddings).
-
-The result: An AI "Incident Response Team" that navigates servers, traces logs, and fixes bugs identically to a human SRE.
-
----
-
-## 🖼️ Application Screenshots
-
-### 📊 Simulation Dashboard
-
-> The core command center. Features live agent terminals, a dual-communication consensus log, and a mathematical performance reward graph plotting investigation confidence.
-
-<div align="center">
-  <img src="./assets/screenshots/Dashboard.png" alt="Simulation Dashboard" width="90%"/>
-</div>
-
----
-
-## 🎛️ Scenario Registry & Core Settings
-
-> The system is architected for instant adaptability — seamlessly switch LLM providers and inject custom threat models entirely through the frontend DOM.
-
-<table>
-  <tr>
-    <td align="center" width="50%">
-      <img src="./assets/screenshots/Scenarios.png" alt="Scenario Browser"/>
-      <br/><b>Scenario Registry</b>
-      <br/><sub>A persistent LocalStorage-backed grid of tactical simulations. Users can dynamically inject custom infrastructure-specific incidents directly into the agent pipeline.</sub>
-    </td>
-    <td align="center" width="50%">
-      <img src="./assets/screenshots/Settings.png" alt="Hardware Configuration"/>
-      <br/><b>Runtime Configuration</b>
-      <br/><sub>Dynamically maps available locally-installed Ollama networks, allowing the user to pair models (e.g., Qwen vs Dolphin-Phi) with fully independent parameters.</sub>
-    </td>
-  </tr>
-</table>
-
----
-
-## 🏗️ System Architecture
-
-```text
-┌─────────────────────────────────────────────────────────────────┐
-│                    CLIENT BROWSER                               │
-│          React SPA (Tailwind + Framer Motion)                   │
-│          localhost:5173                                         │
-└───────────┬─────────────────────────────────┬───────────────────┘
-            │ HTTP (REST)                     │ ws://
-            ▼                                 ▼
-┌─────────────────────────────────────────────────────────────────┐
-│              FASTAPI BACKEND (localhost:7860)                   │
-│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────────────┐    │
-│  │ /config  │ │/scenarios│ │  /reset  │ │  ws:// Simulator │    │
-│  │ Env Sync │ │ DB Cache │ │ Injection│ │  Live Stream Sync│    │
-│  └──────────┘ └──────────┘ └──────────┘ └──────────────────┘    │
-└───────────┬───────────────────────────────────┬─────────────────┘
-            │                                   │
-            ▼                                   ▼
-┌─────────────────────────────────────────────────────────────────┐
-│                  OLLAMA ENGINE / LLM PIPELINE                   │
-│  Agent A (Investigator)   ◄──────►   Agent B (Validator)        │
-│  - Generates Hypotheses              - Challenges Assertions    │
-│  - Runs System Tools                 - Requires Proof           │
-└─────────────────────────────────────────────────────────────────┘
-```
-
----
-
-## 🌐 Execution Environments
-
-NEXUS-AI supports two distinct execution models for agent tools, toggleable via the **Settings** dashboard:
-
-### 1. Simulated Mode (Safe Sandbox)
-*   **Default Mode**: Agents interact with a pre-defined `clue_map` within the scenario YAML.
-*   **No System Impact**: Commands like `read_logs` or `check_service` return mocked data.
-*   **Use Case**: Training, logic validation, and "what-if" analysis without infrastructure risk.
-
-### 2. SSH Lab Node (Real-World Execution)
-*   **Live Connection**: Commands are executed in real-time on a remote Linux server via SSH.
-*   **Autonomous Terminal**: Agents use the `run_terminal_command` tool to browse logs, check systemd status, and inspect real configs.
-*   **Security**: Includes a command blocklist to prevent highly destructive operations (e.g., `rm -rf /`).
-*   **Use Case**: Actual incident response on isolated Lab/Staging nodes.
-
----
-
-## 📐 OpenEnv Specification
-
-NEXUS-AI strictly adheres to the **OpenEnv 1.0** standard for agent-environment interaction.
-
-### 🎮 Action Space
-The environment accepts a typed **NexusAction** (Text-based with structured tool calls).
-- **agent_id**: `string` ("agent_a" or "agent_b")
-- **message**: `string` (The natural language reasoning/communication)
-- **tool_calls**: `List[ToolCall]` (Optional structured calls like `TOOL: read_logs(file='app.log')`)
-- **confidence**: `float` (0.0 - 1.0)
-
-### 🧐 Observation Space
-The environment returns a structured **NexusObservation** summarizing the system state.
-- **scenario_description**: `string` (High-level objective)
-- **scenario_context**: `string` (Background telemetry/environment info)
-- **partner_message**: `string` (The last message from the other agent)
-- **tool_results**: `List[ToolResult]` (Output of any executed system tools)
-- **clues_found**: `List[string]` (Accumulated evidence identified by the Reward Engine)
-- **investigation_stage**: `string` (`investigating`, `narrowing`, `found`, `verified`)
-- **round**: `integer` (Current episode round)
-- **available_tools**: `List[string]` (List of permitted tools for the current mode)
-
-### 📝 Task Registry & Difficulty
-| Task Name | Difficulty | Objective | Grader Method |
-|---|---|---|---|
-| `software-incident` | **Easy** | Fix Nginx 503 rate-limit misconfiguration | State Check: `nginx-proxy.rate_limit` |
-| `business-process-failure` | **Medium** | Resolve inventory stockout logic error | State Check: `stock_threshold` + Red Herring Penalty |
-| `cascade-system-failure` | **Hard** | Fix Postgres connection exhaustion | Multi-Step: Query Termination + Config Update |
-
-### 📈 Baseline Benchmarks
-Validated using `inference.py` (Phi-3-mini & Qwen2.5-1.5B).
-- **Software Incident**: 0.88 / 1.00
-- **Business Process Failure**: 0.72 / 1.00
-- **Cascade System Failure**: 0.48 / 1.00
-
----
-
-## 🧠 The AI Pipeline Deep-Dive
-
-### Step 1: Scenario Injection & Bootstrapping
-```python
-# The EpisodeManager receives the frontend custom scenario JSON
-# Broadcasts 'episode_start' natively over the WebSocket to synchronize the UI
-await broadcast("episode_start", {
-    "scenario": active_scenario,
-    "agent_a_model": settings.AGENT_A_MODEL
-})
-```
-
-### Step 2: Agent Consensus Loop
-```python
-# Agents interact sequentially. The Investigator attempts a solution
-# while the Validator challenges it. Both agents have access to dynamic system execution.
-client, model_name = model_manager.get_client(agent_id)
-stream = await client.chat.completions.create(
-    model=model_name,
-    messages=injected_history,
-    tools=available_tools, # e.g. fix_proposer, run_terminal_command
-    stream=True
-)
-```
-
-### Step 3: Fast GPU Embeddings (Similarity Evaluation)
-```python
-# Heavy CPU blocking is completely bypassed.
-# Semantic embedding computations map strictly into the Ollama GPU pipeline.
-@lru_cache(maxsize=256)
-def get_embedding(text: str) -> List[float]:
-    response = httpx.post("http://localhost:11434/api/embeddings", json={
-        "model": "all-minilm",
-        "prompt": text
-    }, timeout=60.0)
-    return response.json().get("embedding", [])
-```
-
----
-
-## 🛠️ Full Technology Stack
-
-| Layer | Technology | Why |
-|---|---|---|
-| Frontend Framework | React 18 (Vite) | Lightning fast HMR, component isolation |
-| Frontend Styling | Tailwind CSS | Utility-first tactical glassmorphism |
-| Backend Framework | FastAPI | Async Python, explicit endpoint mapping |
-| Transport Layer | WebSockets | Word-by-word streaming across UI boundaries |
-| Local AI Engine | Ollama | Native device acceleration, absolute privacy |
-| Remote Provider | HuggingFace Inference API | Drop-in SaaS alternatives |
-| SSH Connectivity | Paramiko | Secure remote shell execution for Lab Nodes |
-| Data Persistence | LocalStorage & `.env` Injection | Avoids over-architected SQL constraints |
-
----
-
-## 🚀 How to Run This Project (Full Step-by-Step Guide)
-
-### 📋 Prerequisites
-- Python 3.10+
-- Node.js 18+
-- [Ollama](https://ollama.com/) (installed locally for model hosting)
-- **Optional**: A remote Linux VM (Ubuntu/Kali) with SSH enabled for Lab Node mode
-
----
-
-### 1️⃣ Backend Setup (FastAPI / Python)
-
-```bash
-cd backend
-
-# Create and activate virtual environment
-python -m venv venv
-# source venv/bin/activate       # Linux/macOS
-venv\Scripts\activate        # Windows
-
-# Install all dependencies
-pip install -r requirements.txt
-```
-
-#### Start the Backend Engine
-```bash
-# This exposes the core REST API and the WebSocket simulation tunnel
-python main.py
-```
-
----
-
-### 2️⃣ Frontend Setup (React)
-
-Open a **new terminal tab**:
-
-```bash
-cd frontend
-
-# Install Node.js dependencies
-npm install
-
-# Start the Vite development server
-npm run dev
-```
-
-The application is now fully accessible at [http://localhost:5173](http://localhost:5173).
-
----
-
-### 3️⃣ Pulling Models
-
-To run the simulation locally without cloud API keys, you must ensure you pull suitable reasoning models through Ollama:
-
-```bash
-ollama run qwen2.5:3b     # Excellent validator logic footprint
-ollama run dolphin-llama3 # Uncensored investigative assertions
-ollama pull all-minilm    # Mandatory for semantic similarity scoring
-```
-
----
-
-## 🧪 Automated Testing
-NEXUS-AI includes a comprehensive test suite to ensure environment stability and specification compliance.
-
-```bash
-# Run the OpenEnv specification validator
-python openenv_validator.py
-
-# Run unit tests for core logic
-pip install pytest
-pytest tests/
-```
-
----
-
-## 🤝 Authors
-**Developed by: Ashish Menon** & Vector
+---
+title: NEXON-AI
+emoji: 🛡️
+colorFrom: blue
+colorTo: indigo
+sdk: docker
+app_port: 7860
+pinned: false
+---
+
+<!-- LAST_SYNC_VERIFICATION: 2026-04-08 00:07:00 -->
+
+# NEXUS-AI 🌐🛡️
+### Autonomous Incident Investigation Dashboard
+
+<div align="center">
+
+![Python](https://img.shields.io/badge/Python-3.10+-3776AB?style=for-the-badge&logo=python&logoColor=white)
+![FastAPI](https://img.shields.io/badge/FastAPI-0.100+-009688?style=for-the-badge&logo=fastapi&logoColor=white)
+![React](https://img.shields.io/badge/React-18.x-61DAFB?style=for-the-badge&logo=react&logoColor=black)
+![Tailwind](https://img.shields.io/badge/Tailwind_CSS-3.x-38B2AC?style=for-the-badge&logo=tailwind-css&logoColor=white)
+![Ollama](https://img.shields.io/badge/Ollama-Local_LLM-000000?style=for-the-badge&logo=ollama)
+
+**Status:** Active Simulation Pipeline  
+**Architecture:** Real-time WebSockets + Multi-Agent Consensus
+
+</div>
+
+---
+
+## 📖 What is NEXUS-AI?
+
+NEXUS is a next-generation, autonomous dual-agent environment designed to investigate and validate software incidents in real-time. Using a combination of an **Investigator** and a **Validator** agent, NEXUS autonomously forms hypotheses, executes systems tools, evaluates system behavior, and reaches strict consensus on root causes.
+
+Traditional manual debugging requires extensive context-switching and tool fatigue. NEXUS solves this through:
+1. **Dual-Agent Autonomy**: Two specialized models communicating word-by-word via WebSockets.
+2. **Dynamic Tool Execution**: Fully integrated system terminals allowing agents to run sandboxed validation scripts.
+3. **Semantic Reward Engine**: Evaluates conversational drift mathematically (using native GPU embeddings).
+
+The result: An AI "Incident Response Team" that navigates servers, traces logs, and fixes bugs identically to a human SRE.
+
+---
+
+## 🖼️ Application Screenshots
+
+### 📊 Simulation Dashboard
+
+> The core command center. Features live agent terminals, a dual-communication consensus log, and a mathematical performance reward graph plotting investigation confidence.
+
+<div align="center">
+  <img src="./assets/screenshots/Dashboard.png" alt="Simulation Dashboard" width="90%"/>
+</div>
+
+---
+
+## 🎛️ Scenario Registry & Core Settings
+
+> The system is architected for instant adaptability — seamlessly switch LLM providers and inject custom threat models entirely through the frontend DOM.
+
+<table>
+  <tr>
+    <td align="center" width="50%">
+      <img src="./assets/screenshots/Scenarios.png" alt="Scenario Browser"/>
+      <br/><b>Scenario Registry</b>
+      <br/><sub>A persistent LocalStorage-backed grid of tactical simulations. Users can dynamically inject custom infrastructure-specific incidents directly into the agent pipeline.</sub>
+    </td>
+    <td align="center" width="50%">
+      <img src="./assets/screenshots/Settings.png" alt="Hardware Configuration"/>
+      <br/><b>Runtime Configuration</b>
+      <br/><sub>Dynamically maps available locally-installed Ollama networks, allowing the user to pair models (e.g., Qwen vs Dolphin-Phi) with fully independent parameters.</sub>
+    </td>
+  </tr>
+</table>
+
+---
+
+## 🏗️ System Architecture
+
+```text
+┌─────────────────────────────────────────────────────────────────┐
+│                    CLIENT BROWSER                               │
+│          React SPA (Tailwind + Framer Motion)                   │
+│          localhost:5173                                         │
+└───────────┬─────────────────────────────────┬───────────────────┘
+            │ HTTP (REST)                     │ ws://
+            ▼                                 ▼
+┌─────────────────────────────────────────────────────────────────┐
+│              FASTAPI BACKEND (localhost:7860)                   │
+│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────────────┐    │
+│  │ /config  │ │/scenarios│ │  /reset  │ │  ws:// Simulator │    │
+│  │ Env Sync │ │ DB Cache │ │ Injection│ │  Live Stream Sync│    │
+│  └──────────┘ └──────────┘ └──────────┘ └──────────────────┘    │
+└───────────┬───────────────────────────────────┬─────────────────┘
+            │                                   │
+            ▼                                   ▼
+┌─────────────────────────────────────────────────────────────────┐
+│                  OLLAMA ENGINE / LLM PIPELINE                   │
+│  Agent A (Investigator)   ◄──────►   Agent B (Validator)        │
+│  - Generates Hypotheses              - Challenges Assertions    │
+│  - Runs System Tools                 - Requires Proof           │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## 🌐 Execution Environments
+
+NEXUS-AI supports two distinct execution models for agent tools, toggleable via the **Settings** dashboard:
+
+### 1. Simulated Mode (Safe Sandbox)
+*   **Default Mode**: Agents interact with a pre-defined `clue_map` within the scenario YAML.
+*   **No System Impact**: Commands like `read_logs` or `check_service` return mocked data.
+*   **Use Case**: Training, logic validation, and "what-if" analysis without infrastructure risk.
+
+### 2. SSH Lab Node (Real-World Execution)
+*   **Live Connection**: Commands are executed in real-time on a remote Linux server via SSH.
+*   **Autonomous Terminal**: Agents use the `run_terminal_command` tool to browse logs, check systemd status, and inspect real configs.
+*   **Security**: Includes a command blocklist to prevent highly destructive operations (e.g., `rm -rf /`).
+*   **Use Case**: Actual incident response on isolated Lab/Staging nodes.
+
+---
+
+## 📐 OpenEnv Specification
+
+NEXUS-AI strictly adheres to the **OpenEnv 1.0** standard for agent-environment interaction.
+
+### 🎮 Action Space
+The environment accepts a typed **NexusAction** (Text-based with structured tool calls).
+- **agent_id**: `string` ("agent_a" or "agent_b")
+- **message**: `string` (The natural language reasoning/communication)
+- **tool_calls**: `List[ToolCall]` (Optional structured calls like `TOOL: read_logs(file='app.log')`)
+- **confidence**: `float` (0.0 - 1.0)
+
+### 🧐 Observation Space
+The environment returns a structured **NexusObservation** summarizing the system state.
+- **scenario_description**: `string` (High-level objective)
+- **scenario_context**: `string` (Background telemetry/environment info)
+- **partner_message**: `string` (The last message from the other agent)
+- **tool_results**: `List[ToolResult]` (Output of any executed system tools)
+- **clues_found**: `List[string]` (Accumulated evidence identified by the Reward Engine)
+- **investigation_stage**: `string` (`investigating`, `narrowing`, `found`, `verified`)
+- **round**: `integer` (Current episode round)
+- **available_tools**: `List[string]` (List of permitted tools for the current mode)
+
+### 📝 Task Registry & Difficulty
+| Task Name | Difficulty | Objective | Grader Method |
+|---|---|---|---|
+| `software-incident` | **Easy** | Fix Nginx 503 rate-limit misconfiguration | State Check: `nginx-proxy.rate_limit` |
+| `business-process-failure` | **Medium** | Resolve inventory stockout logic error | State Check: `stock_threshold` + Red Herring Penalty |
+| `cascade-system-failure` | **Hard** | Fix Postgres connection exhaustion | Multi-Step: Query Termination + Config Update |
+
+### 📈 Baseline Benchmarks
+Validated using `inference.py` (Phi-3-mini & Qwen2.5-1.5B).
+- **Software Incident**: 0.88 / 1.00
+- **Business Process Failure**: 0.72 / 1.00
+- **Cascade System Failure**: 0.48 / 1.00
+
+---
+
+## 🧠 The AI Pipeline Deep-Dive
+
+### Step 1: Scenario Injection & Bootstrapping
+```python
+# The EpisodeManager receives the frontend custom scenario JSON
+# Broadcasts 'episode_start' natively over the WebSocket to synchronize the UI
+await broadcast("episode_start", {
+    "scenario": active_scenario,
+    "agent_a_model": settings.AGENT_A_MODEL
+})
+```
+
+### Step 2: Agent Consensus Loop
+```python
+# Agents interact sequentially. The Investigator attempts a solution
+# while the Validator challenges it. Both agents have access to dynamic system execution.
+client, model_name = model_manager.get_client(agent_id)
+stream = await client.chat.completions.create(
+    model=model_name,
+    messages=injected_history,
+    tools=available_tools, # e.g. fix_proposer, run_terminal_command
+    stream=True
+)
+```
+
+### Step 3: Fast GPU Embeddings (Similarity Evaluation)
+```python
+# Heavy CPU blocking is completely bypassed.
+# Semantic embedding computations map strictly into the Ollama GPU pipeline.
+@lru_cache(maxsize=256)
+def get_embedding(text: str) -> List[float]:
+    response = httpx.post("http://localhost:11434/api/embeddings", json={
+        "model": "all-minilm",
+        "prompt": text
+    }, timeout=60.0)
+    return response.json().get("embedding", [])
+```
+
+---
+
+## 🛠️ Full Technology Stack
+
+| Layer | Technology | Why |
+|---|---|---|
+| Frontend Framework | React 18 (Vite) | Lightning fast HMR, component isolation |
+| Frontend Styling | Tailwind CSS | Utility-first tactical glassmorphism |
+| Backend Framework | FastAPI | Async Python, explicit endpoint mapping |
+| Transport Layer | WebSockets | Word-by-word streaming across UI boundaries |
+| Local AI Engine | Ollama | Native device acceleration, absolute privacy |
+| Remote Provider | HuggingFace Inference API | Drop-in SaaS alternatives |
+| SSH Connectivity | Paramiko | Secure remote shell execution for Lab Nodes |
+| Data Persistence | LocalStorage & `.env` Injection | Avoids over-architected SQL constraints |
+
+---
+
+## 🚀 How to Run This Project (Full Step-by-Step Guide)
+
+### 📋 Prerequisites
+- Python 3.10+
+- Node.js 18+
+- [Ollama](https://ollama.com/) (installed locally for model hosting)
+- **Optional**: A remote Linux VM (Ubuntu/Kali) with SSH enabled for Lab Node mode
+
+---
+
+### 1️⃣ Backend Setup (FastAPI / Python)
+
+```bash
+cd backend
+
+# Create and activate virtual environment
+python -m venv venv
+# source venv/bin/activate       # Linux/macOS
+venv\Scripts\activate        # Windows
+
+# Install all dependencies
+pip install -r requirements.txt
+```
+
+#### Start the Backend Engine
+```bash
+# This exposes the core REST API and the WebSocket simulation tunnel
+python main.py
+```
+
+---
+
+### 2️⃣ Frontend Setup (React)
+
+Open a **new terminal tab**:
+
+```bash
+cd frontend
+
+# Install Node.js dependencies
+npm install
+
+# Start the Vite development server
+npm run dev
+```
+
+The application is now fully accessible at [http://localhost:5173](http://localhost:5173).
+
+---
+
+### 3️⃣ Pulling Models
+
+To run the simulation locally without cloud API keys, you must ensure you pull suitable reasoning models through Ollama:
+
+```bash
+ollama run qwen2.5:3b     # Excellent validator logic footprint
+ollama run dolphin-llama3 # Uncensored investigative assertions
+ollama pull all-minilm    # Mandatory for semantic similarity scoring
+```
+
+---
+
+## 🧪 Automated Testing
+NEXUS-AI includes a comprehensive test suite to ensure environment stability and specification compliance.
+
+```bash
+# Run the OpenEnv specification validator
+python openenv_validator.py
+
+# Run unit tests for core logic
+pip install pytest
+pytest tests/
+```
+
+---
+
+## 🤝 Authors
+**Developed by: Ashish Menon** & Vector

SYNC_VERIFICATION_0047.txt

@@ -1,5 +1,5 @@
-This file is a marker to verify that the synchronization between the local environment and the remote repositories (GitHub and Hugging Face) is functioning correctly.
-
-Timestamp: 2026-04-08 00:47:00
-Commit SHA: 4f14584 (previous)
-Status: FORCED_SYNC_ACTIVE
+This file is a marker to verify that the synchronization between the local environment and the remote repositories (GitHub and Hugging Face) is functioning correctly.
+
+Timestamp: 2026-04-08 00:47:00
+Commit SHA: 4f14584 (previous)
+Status: FORCED_SYNC_ACTIVE

backend/config.py

@@ -1,75 +1,75 @@
-import os
-from pathlib import Path
-from dotenv import load_dotenv
-
-BASE_DIR = Path(__file__).resolve().parent
-ROOT_DIR = BASE_DIR.parent
-
-# Load environment variables, checking both backend/ and project root
-if (BASE_DIR / ".env").exists():
-    load_dotenv(BASE_DIR / ".env")
-elif (ROOT_DIR / ".env").exists():
-    load_dotenv(ROOT_DIR / ".env")
-elif (ROOT_DIR / "default.env").exists():
-    load_dotenv(ROOT_DIR / "default.env")
-else:
-    load_dotenv() # Fallback to standard search
-
-class Settings:
-    # OLLAMA
-    OLLAMA_BASE_URL = os.getenv("OLLAMA_BASE_URL", "http://localhost:11434/v1")
-    OLLAMA_API_KEY = os.getenv("OLLAMA_API_KEY", "ollama")
-
-    # AGENTS
-    AGENT_A_MODEL = os.getenv("AGENT_A_MODEL", "")
-    AGENT_B_MODEL = os.getenv("AGENT_B_MODEL", "")
-    AGENT_A_PROVIDER = os.getenv("AGENT_A_PROVIDER", "ollama")
-    AGENT_B_PROVIDER = os.getenv("AGENT_B_PROVIDER", "ollama")
-    AGENT_A_ROLE = os.getenv("AGENT_A_ROLE", "INVESTIGATOR")
-    AGENT_B_ROLE = os.getenv("AGENT_B_ROLE", "VALIDATOR")
-    AGENT_A_SYSTEM_PROMPT = os.getenv("AGENT_A_SYSTEM_PROMPT", "")
-    AGENT_B_SYSTEM_PROMPT = os.getenv("AGENT_B_SYSTEM_PROMPT", "")
-    AGENT_A_TEMPERATURE = float(os.getenv("AGENT_A_TEMPERATURE", "0.8"))
-    AGENT_B_TEMPERATURE = float(os.getenv("AGENT_B_TEMPERATURE", "0.6"))
-    AGENT_A_MAX_TOKENS = int(os.getenv("AGENT_A_MAX_TOKENS", "300"))
-    AGENT_B_MAX_TOKENS = int(os.getenv("AGENT_B_MAX_TOKENS", "300"))
-    # EXECUTION ENVIRONMENT
-    EXECUTION_MODE = os.getenv("EXECUTION_MODE", "simulated")
-    SSH_HOST = os.getenv("SSH_HOST", "")
-    SSH_PORT = int(os.getenv("SSH_PORT", "22"))
-    SSH_USER = os.getenv("SSH_USER", "")
-    SSH_PASSWORD = os.getenv("SSH_PASSWORD", "")
-
-    # HUGGINGFACE
-    API_KEY = os.getenv("API_KEY", "ollama")
-    OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "")
-    HF_TOKEN = os.getenv("HF_TOKEN", "")
-    HF_INFERENCE_URL = os.getenv("HF_INFERENCE_URL", "https://router.huggingface.co/v1")
-
-    # OPENROUTER
-    OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY", "")
-    OPENROUTER_BASE_URL = os.getenv("OPENROUTER_BASE_URL", "https://openrouter.ai/api/v1")
-
-    # SERVER
-    HOST = os.getenv("HOST", "0.0.0.0")
-    PORT = int(os.getenv("PORT", "7860"))
-    DEBUG = os.getenv("DEBUG", "true").lower() in ("true", "1", "yes")
-    ENVIRONMENT = os.getenv("ENVIRONMENT", "local")
-
-    # EPISODE
-    MAX_STEPS = int(os.getenv("MAX_STEPS", "1000"))
-    MAX_EPISODE_TIME_SECONDS = int(os.getenv("MAX_EPISODE_TIME_SECONDS", "1200"))
-    SUCCESS_SCORE_THRESHOLD = float(os.getenv("SUCCESS_SCORE_THRESHOLD", "0.5"))
-
-    # MCP TOOL SERVER
-    MCP_SERVER_PORT = int(os.getenv("MCP_SERVER_PORT", "8001"))
-    MCP_SERVER_URL = os.getenv("MCP_SERVER_URL", "http://localhost:8001")
-
-    # CUSTOM MODEL
-    CUSTOM_MODEL_ENABLED = os.getenv("CUSTOM_MODEL_ENABLED", "false").lower() in ("true", "1", "yes")
-    CUSTOM_MODEL_BASE_URL = os.getenv("CUSTOM_MODEL_BASE_URL", "")
-    CUSTOM_MODEL_API_KEY = os.getenv("CUSTOM_MODEL_API_KEY", "")
-    CUSTOM_MODEL_NAME = os.getenv("CUSTOM_MODEL_NAME", "")
-    CUSTOM_MODEL_AGENT = os.getenv("CUSTOM_MODEL_AGENT", "")
-
-settings = Settings()
+import os
+from pathlib import Path
+from dotenv import load_dotenv
+
+BASE_DIR = Path(__file__).resolve().parent
+ROOT_DIR = BASE_DIR.parent
+
+# Load environment variables, checking both backend/ and project root
+if (BASE_DIR / ".env").exists():
+    load_dotenv(BASE_DIR / ".env")
+elif (ROOT_DIR / ".env").exists():
+    load_dotenv(ROOT_DIR / ".env")
+elif (ROOT_DIR / "default.env").exists():
+    load_dotenv(ROOT_DIR / "default.env")
+else:
+    load_dotenv() # Fallback to standard search
+
+class Settings:
+    # OLLAMA
+    OLLAMA_BASE_URL = os.getenv("OLLAMA_BASE_URL", "http://localhost:11434/v1")
+    OLLAMA_API_KEY = os.getenv("OLLAMA_API_KEY", "ollama")
+
+    # AGENTS
+    AGENT_A_MODEL = os.getenv("AGENT_A_MODEL", "")
+    AGENT_B_MODEL = os.getenv("AGENT_B_MODEL", "")
+    AGENT_A_PROVIDER = os.getenv("AGENT_A_PROVIDER", "ollama")
+    AGENT_B_PROVIDER = os.getenv("AGENT_B_PROVIDER", "ollama")
+    AGENT_A_ROLE = os.getenv("AGENT_A_ROLE", "INVESTIGATOR")
+    AGENT_B_ROLE = os.getenv("AGENT_B_ROLE", "VALIDATOR")
+    AGENT_A_SYSTEM_PROMPT = os.getenv("AGENT_A_SYSTEM_PROMPT", "")
+    AGENT_B_SYSTEM_PROMPT = os.getenv("AGENT_B_SYSTEM_PROMPT", "")
+    AGENT_A_TEMPERATURE = float(os.getenv("AGENT_A_TEMPERATURE", "0.8"))
+    AGENT_B_TEMPERATURE = float(os.getenv("AGENT_B_TEMPERATURE", "0.6"))
+    AGENT_A_MAX_TOKENS = int(os.getenv("AGENT_A_MAX_TOKENS", "300"))
+    AGENT_B_MAX_TOKENS = int(os.getenv("AGENT_B_MAX_TOKENS", "300"))
+    # EXECUTION ENVIRONMENT
+    EXECUTION_MODE = os.getenv("EXECUTION_MODE", "simulated")
+    SSH_HOST = os.getenv("SSH_HOST", "")
+    SSH_PORT = int(os.getenv("SSH_PORT", "22"))
+    SSH_USER = os.getenv("SSH_USER", "")
+    SSH_PASSWORD = os.getenv("SSH_PASSWORD", "")
+
+    # HUGGINGFACE
+    API_KEY = os.getenv("API_KEY", "ollama")
+    OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "")
+    HF_TOKEN = os.getenv("HF_TOKEN", "")
+    HF_INFERENCE_URL = os.getenv("HF_INFERENCE_URL", "https://router.huggingface.co/v1")
+
+    # OPENROUTER
+    OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY", "")
+    OPENROUTER_BASE_URL = os.getenv("OPENROUTER_BASE_URL", "https://openrouter.ai/api/v1")
+
+    # SERVER
+    HOST = os.getenv("HOST", "0.0.0.0")
+    PORT = int(os.getenv("PORT", "7860"))
+    DEBUG = os.getenv("DEBUG", "true").lower() in ("true", "1", "yes")
+    ENVIRONMENT = os.getenv("ENVIRONMENT", "local")
+
+    # EPISODE
+    MAX_STEPS = int(os.getenv("MAX_STEPS", "1000"))
+    MAX_EPISODE_TIME_SECONDS = int(os.getenv("MAX_EPISODE_TIME_SECONDS", "1200"))
+    SUCCESS_SCORE_THRESHOLD = float(os.getenv("SUCCESS_SCORE_THRESHOLD", "0.5"))
+
+    # MCP TOOL SERVER
+    MCP_SERVER_PORT = int(os.getenv("MCP_SERVER_PORT", "8001"))
+    MCP_SERVER_URL = os.getenv("MCP_SERVER_URL", "http://localhost:8001")
+
+    # CUSTOM MODEL
+    CUSTOM_MODEL_ENABLED = os.getenv("CUSTOM_MODEL_ENABLED", "false").lower() in ("true", "1", "yes")
+    CUSTOM_MODEL_BASE_URL = os.getenv("CUSTOM_MODEL_BASE_URL", "")
+    CUSTOM_MODEL_API_KEY = os.getenv("CUSTOM_MODEL_API_KEY", "")
+    CUSTOM_MODEL_NAME = os.getenv("CUSTOM_MODEL_NAME", "")
+    CUSTOM_MODEL_AGENT = os.getenv("CUSTOM_MODEL_AGENT", "")
+
+settings = Settings()

backend/core/environment.py

@@ -1,160 +1,160 @@
-import json
-from typing import Tuple, Dict
-
-from scenarios.scenario_loader import scenario_loader
-from core.state_manager import EpisodeState
-from core.reward_engine import compute_reward
-from core.agent_runner import AgentRunner
-from scenarios.graders.easy_grader import EasyGrader
-from scenarios.graders.medium_grader import MediumGrader
-from scenarios.graders.hard_grader import HardGrader
-from api.schemas.action import NexusAction
-from api.schemas.observation import NexusObservation, ToolResult
-from config import settings
-import statistics
-
-SIMULATED_TOOLS = ["read_logs", "check_config", "query_database", "check_service_status", "run_diagnostic", "update_config", "restart_service", "propose_fix", "verify_fix", "submit_resolution"]
-SSH_TOOLS = ["run_terminal_command", "propose_fix", "verify_fix", "submit_resolution"]
-
-class NexusEnvironment:
-    def __init__(self):
-        self.runner = AgentRunner()
-        self.active_episode = None
-        self.active_scenario = None
-        
-        self.graders = {
-            "easy": EasyGrader(),
-            "medium": MediumGrader(),
-            "hard": HardGrader()
-        }
-
-    async def reset(self, task: str = "software-incident", scenario_id: str = None, custom_scenario: dict = None, seed: int = None, max_steps: int = None) -> NexusObservation:
-        # Determine difficulty from task
-        valid_tasks = ["software-incident", "business-process-failure", "cascade-system-failure"]
-        if task not in valid_tasks and not custom_scenario and not scenario_id:
-            raise ValueError(f"Invalid task name: {task}")
-            
-        difficulty = "easy"
-        if task == "business-process-failure":
-            difficulty = "medium"
-        elif task == "cascade-system-failure":
-            difficulty = "hard"
-            
-        if custom_scenario:
-            scenario = custom_scenario
-            scenario["id"] = scenario.get("id", "custom-1")
-            scenario["description"] = scenario.get("description", "Custom imported scenario.")
-            scenario["context"] = scenario.get("context", "Custom uploaded environment.")
-            if "difficulty" in scenario:
-                 difficulty = scenario["difficulty"].lower()
-        elif scenario_id:
-            scenario = scenario_loader.get_scenario(scenario_id)
-        else:
-            scenarios = scenario_loader.get_scenarios_by_difficulty(difficulty)
-            if not scenarios:
-                raise ValueError(f"No scenarios found for difficulty {difficulty}")
-            import random
-            if seed is not None:
-                random.seed(seed)
-            scenario = random.choice(scenarios)
-            
-        self.active_scenario = scenario
-        self.active_episode = EpisodeState(
-            scenario_id=scenario["id"],
-            task=task,
-            difficulty=difficulty,
-            max_rounds=max_steps if max_steps is not None else settings.MAX_STEPS,
-            scenario_data=scenario
-        )
-        
-        available_tools = SSH_TOOLS if settings.EXECUTION_MODE == "ssh" else SIMULATED_TOOLS
-        obs = NexusObservation(
-            partner_message="",
-            tool_results=[],
-            system_state={},
-            investigation_stage="investigating",
-            round=1,
-            available_tools=available_tools,
-            clues_found=[],
-            scenario_description=scenario["description"],
-            scenario_context=scenario["context"]
-        )
-        return obs
-        
-    async def step(self, action: NexusAction) -> Tuple[NexusObservation, float, bool, dict]:
-        if not self.active_episode:
-            raise ValueError("Environment must be reset before calling step")
-            
-        ep = self.active_episode
-        sc = self.active_scenario
-        
-        # 1. Add agent message to state
-        ep.add_message(action.agent_id, action.message)
-        
-        # 2. Execute tools
-        tool_results_data = await self.runner.execute_tool_calls(action.tool_calls, sc, ep.current_round, ep)
-        
-        # Process tool clues
-        tool_results_objs = []
-        for tr in tool_results_data:
-            if "status: degraded" in tr['result'].lower() or "error" in tr['result'].lower() or "anomaly" in tr['result'].lower() or "warning" in tr['result'].lower() or tr['tool_name'] == 'propose_fix' or tr['tool_name'] == 'verify_fix':
-                ep.add_clue(tr['result'])
-            tool_results_objs.append(ToolResult(**tr))
-            
-        # 3. Compute semantic reward dynamically
-        reward, breakdown = compute_reward(action.message, action.tool_calls, tool_results_data, ep, sc)
-        
-        # Stop when resolution submitted or max steps taken
-        if ep.fix_verified or ep.steps_taken >= ep.max_rounds:
-            ep.done = True
-            
-            # If they maxed out without resolving, inject a synthetic report so the UI doesn't look broken
-            if not ep.fix_verified:
-                ep.add_tool_call("submit_resolution", {
-                    "root_cause_service": "UNRESOLVED",
-                    "root_cause_description": "Investigation terminated: Maximum round limit reached without agent consensus.",
-                    "fix_applied": "No fix was submitted."
-                })
-            
-            # Hybrid Final Scorer: Combine objective grader results with semantic reward history
-            grader = self.graders.get(ep.difficulty, self.graders["easy"])
-            grader_score = grader.grade(ep, sc)
-            
-            # Use average step reward as the semantic component (0.0 - 1.0)
-            avg_semantic = statistics.mean(ep.reward_history) if ep.reward_history else 0.0
-            
-            # Weighted average: Grader (Objective) 60% + Semantic (Quality) 40%
-            # If the grader score is 1.0 (perfect fix), we lean more into the objective truth.
-            if grader_score >= 0.90:
-                final_score = grader_score * 0.8 + avg_semantic * 0.2
-            else:
-                final_score = grader_score * 0.6 + avg_semantic * 0.4
-            
-            final_score = round(max(0.0, min(1.0, final_score)), 4)
-            
-            info = {
-                "breakdown": {**breakdown, "semantic_avg": round(avg_semantic, 4), "objective_score": grader_score},
-                "final_score": final_score,
-                "success": (final_score >= settings.SUCCESS_SCORE_THRESHOLD) or (ep.fix_verified and grader_score > 0)
-            }
-        else:
-            info = {"breakdown": breakdown}
-
-        obs = NexusObservation(
-            partner_message=action.message,
-            tool_results=tool_results_objs,
-            system_state={"total_tools_run": len(ep.tool_calls_made)},
-            investigation_stage=ep.investigation_stage,
-            round=ep.current_round,
-            available_tools=SSH_TOOLS if settings.EXECUTION_MODE == "ssh" else SIMULATED_TOOLS,
-            clues_found=ep.clues_found,
-            scenario_description=sc["description"],
-            scenario_context=sc["context"]
-        )
-        
-        return obs, reward, ep.done, info
-
-    def state(self):
-        if not self.active_episode:
-            return None
-        return self.active_episode.to_pydantic()
+import json
+from typing import Tuple, Dict
+
+from scenarios.scenario_loader import scenario_loader
+from core.state_manager import EpisodeState
+from core.reward_engine import compute_reward
+from core.agent_runner import AgentRunner
+from scenarios.graders.easy_grader import EasyGrader
+from scenarios.graders.medium_grader import MediumGrader
+from scenarios.graders.hard_grader import HardGrader
+from api.schemas.action import NexusAction
+from api.schemas.observation import NexusObservation, ToolResult
+from config import settings
+import statistics
+
+SIMULATED_TOOLS = ["read_logs", "check_config", "query_database", "check_service_status", "run_diagnostic", "update_config", "restart_service", "propose_fix", "verify_fix", "submit_resolution"]
+SSH_TOOLS = ["run_terminal_command", "propose_fix", "verify_fix", "submit_resolution"]
+
+class NexusEnvironment:
+    def __init__(self):
+        self.runner = AgentRunner()
+        self.active_episode = None
+        self.active_scenario = None
+        
+        self.graders = {
+            "easy": EasyGrader(),
+            "medium": MediumGrader(),
+            "hard": HardGrader()
+        }
+
+    async def reset(self, task: str = "software-incident", scenario_id: str = None, custom_scenario: dict = None, seed: int = None, max_steps: int = None) -> NexusObservation:
+        # Determine difficulty from task
+        valid_tasks = ["software-incident", "business-process-failure", "cascade-system-failure"]
+        if task not in valid_tasks and not custom_scenario and not scenario_id:
+            raise ValueError(f"Invalid task name: {task}")
+            
+        difficulty = "easy"
+        if task == "business-process-failure":
+            difficulty = "medium"
+        elif task == "cascade-system-failure":
+            difficulty = "hard"
+            
+        if custom_scenario:
+            scenario = custom_scenario
+            scenario["id"] = scenario.get("id", "custom-1")
+            scenario["description"] = scenario.get("description", "Custom imported scenario.")
+            scenario["context"] = scenario.get("context", "Custom uploaded environment.")
+            if "difficulty" in scenario:
+                 difficulty = scenario["difficulty"].lower()
+        elif scenario_id:
+            scenario = scenario_loader.get_scenario(scenario_id)
+        else:
+            scenarios = scenario_loader.get_scenarios_by_difficulty(difficulty)
+            if not scenarios:
+                raise ValueError(f"No scenarios found for difficulty {difficulty}")
+            import random
+            if seed is not None:
+                random.seed(seed)
+            scenario = random.choice(scenarios)
+            
+        self.active_scenario = scenario
+        self.active_episode = EpisodeState(
+            scenario_id=scenario["id"],
+            task=task,
+            difficulty=difficulty,
+            max_rounds=max_steps if max_steps is not None else settings.MAX_STEPS,
+            scenario_data=scenario
+        )
+        
+        available_tools = SSH_TOOLS if settings.EXECUTION_MODE == "ssh" else SIMULATED_TOOLS
+        obs = NexusObservation(
+            partner_message="",
+            tool_results=[],
+            system_state={},
+            investigation_stage="investigating",
+            round=1,
+            available_tools=available_tools,
+            clues_found=[],
+            scenario_description=scenario["description"],
+            scenario_context=scenario["context"]
+        )
+        return obs
+        
+    async def step(self, action: NexusAction) -> Tuple[NexusObservation, float, bool, dict]:
+        if not self.active_episode:
+            raise ValueError("Environment must be reset before calling step")
+            
+        ep = self.active_episode
+        sc = self.active_scenario
+        
+        # 1. Add agent message to state
+        ep.add_message(action.agent_id, action.message)
+        
+        # 2. Execute tools
+        tool_results_data = await self.runner.execute_tool_calls(action.tool_calls, sc, ep.current_round, ep)
+        
+        # Process tool clues
+        tool_results_objs = []
+        for tr in tool_results_data:
+            if "status: degraded" in tr['result'].lower() or "error" in tr['result'].lower() or "anomaly" in tr['result'].lower() or "warning" in tr['result'].lower() or tr['tool_name'] == 'propose_fix' or tr['tool_name'] == 'verify_fix':
+                ep.add_clue(tr['result'])
+            tool_results_objs.append(ToolResult(**tr))
+            
+        # 3. Compute semantic reward dynamically
+        reward, breakdown = compute_reward(action.message, action.tool_calls, tool_results_data, ep, sc)
+        
+        # Stop when resolution submitted or max steps taken
+        if ep.fix_verified or ep.steps_taken >= ep.max_rounds:
+            ep.done = True
+            
+            # If they maxed out without resolving, inject a synthetic report so the UI doesn't look broken
+            if not ep.fix_verified:
+                ep.add_tool_call("submit_resolution", {
+                    "root_cause_service": "UNRESOLVED",
+                    "root_cause_description": "Investigation terminated: Maximum round limit reached without agent consensus.",
+                    "fix_applied": "No fix was submitted."
+                })
+            
+            # Hybrid Final Scorer: Combine objective grader results with semantic reward history
+            grader = self.graders.get(ep.difficulty, self.graders["easy"])
+            grader_score = grader.grade(ep, sc)
+            
+            # Use average step reward as the semantic component (0.0 - 1.0)
+            avg_semantic = statistics.mean(ep.reward_history) if ep.reward_history else 0.0
+            
+            # Weighted average: Grader (Objective) 60% + Semantic (Quality) 40%
+            # If the grader score is 1.0 (perfect fix), we lean more into the objective truth.
+            if grader_score >= 0.90:
+                final_score = grader_score * 0.8 + avg_semantic * 0.2
+            else:
+                final_score = grader_score * 0.6 + avg_semantic * 0.4
+            
+            final_score = round(max(0.0, min(1.0, final_score)), 4)
+            
+            info = {
+                "breakdown": {**breakdown, "semantic_avg": round(avg_semantic, 4), "objective_score": grader_score},
+                "final_score": final_score,
+                "success": (final_score >= settings.SUCCESS_SCORE_THRESHOLD) or (ep.fix_verified and grader_score > 0)
+            }
+        else:
+            info = {"breakdown": breakdown}
+
+        obs = NexusObservation(
+            partner_message=action.message,
+            tool_results=tool_results_objs,
+            system_state={"total_tools_run": len(ep.tool_calls_made)},
+            investigation_stage=ep.investigation_stage,
+            round=ep.current_round,
+            available_tools=SSH_TOOLS if settings.EXECUTION_MODE == "ssh" else SIMULATED_TOOLS,
+            clues_found=ep.clues_found,
+            scenario_description=sc["description"],
+            scenario_context=sc["context"]
+        )
+        
+        return obs, reward, ep.done, info
+
+    def state(self):
+        if not self.active_episode:
+            return None
+        return self.active_episode.to_pydantic()

backend/core/episode_manager.py

@@ -1,95 +1,95 @@
-import asyncio
-from core.environment import NexusEnvironment
-from api.routes.websocket import broadcast
-
-class EpisodeManager:
-    """Manages active episodes and coordinates the WebSocket emissions."""
-    def __init__(self):
-        self.env = NexusEnvironment()
-        self.is_paused = False
-        self.simulation_task = None
-
-    async def reset(self, task: str, custom_scenario: dict = None, seed: int = None, max_steps: int = None, broadcast_episode: bool = True):
-        # Cancel any active simulation loop
-        if hasattr(self, 'simulation_task') and self.simulation_task and not self.simulation_task.done():
-            self.simulation_task.cancel()
-            try:
-                await self.simulation_task
-            except asyncio.CancelledError:
-                pass
-            self.simulation_task = None
-
-        obs = await self.env.reset(task=task, custom_scenario=custom_scenario, seed=seed, max_steps=max_steps)
-        
-        if broadcast_episode:
-            # Broadcast episode_start
-            sc_safe = self.env.active_scenario.copy()
-            if "root_cause" in sc_safe: del sc_safe["root_cause"]
-            if "correct_fix" in sc_safe: del sc_safe["correct_fix"]
-            if "clue_map" in sc_safe: del sc_safe["clue_map"]
-            
-            from config import settings
-            await broadcast("episode_start", {
-                "episode_id": self.env.active_episode.episode_id,
-                "scenario": sc_safe,
-                "task": task,
-                "difficulty": self.env.active_episode.difficulty,
-                "agent_a_model": settings.AGENT_A_MODEL,
-                "agent_b_model": settings.AGENT_B_MODEL
-            })
+import asyncio
+from core.environment import NexusEnvironment
+from api.routes.websocket import broadcast
+
+class EpisodeManager:
+    """Manages active episodes and coordinates the WebSocket emissions."""
+    def __init__(self):
+        self.env = NexusEnvironment()
+        self.is_paused = False
+        self.simulation_task = None
+
+    async def reset(self, task: str, custom_scenario: dict = None, seed: int = None, max_steps: int = None, broadcast_episode: bool = True):
+        # Cancel any active simulation loop
+        if hasattr(self, 'simulation_task') and self.simulation_task and not self.simulation_task.done():
+            self.simulation_task.cancel()
+            try:
+                await self.simulation_task
+            except asyncio.CancelledError:
+                pass
+            self.simulation_task = None
+
+        obs = await self.env.reset(task=task, custom_scenario=custom_scenario, seed=seed, max_steps=max_steps)
+        
+        if broadcast_episode:
+            # Broadcast episode_start
+            sc_safe = self.env.active_scenario.copy()
+            if "root_cause" in sc_safe: del sc_safe["root_cause"]
+            if "correct_fix" in sc_safe: del sc_safe["correct_fix"]
+            if "clue_map" in sc_safe: del sc_safe["clue_map"]
+            
+            from config import settings
+            await broadcast("episode_start", {
+                "episode_id": self.env.active_episode.episode_id,
+                "scenario": sc_safe,
+                "task": task,
+                "difficulty": self.env.active_episode.difficulty,
+                "agent_a_model": settings.AGENT_A_MODEL,
+                "agent_b_model": settings.AGENT_B_MODEL
+            })
         return obs
-
-    async def step(self, action):
-        obs, reward, done, info = await self.env.step(action)
-        
-        # Broadcast agent message
-        await broadcast("agent_message", {
-            "agent_id": action.agent_id,
-            "message": action.message,
-            "step": self.env.active_episode.steps_taken
-        })
-        
-        # Broadcast tool calls
-        for tc in action.tool_calls:
-            await broadcast("tool_call", {
-                "agent_id": action.agent_id,
-                "tool_name": tc.tool_name,
-                "params": tc.params,
-                "step": self.env.active_episode.steps_taken
-            })
-            
-        # Broadcast tool results
-        for tr in obs.tool_results:
-            await broadcast("tool_result", {
-                "tool_name": tr.tool_name,
-                "result": tr.result,
-                "success": tr.success,
-                "step": self.env.active_episode.steps_taken
-            })
-            
-        # Broadcast reward
-        await broadcast("reward_update", {
-            "agent_id": action.agent_id,
-            "reward": reward,
-            "breakdown": info.get("breakdown", {}),
-            "cumulative": self.env.active_episode.cumulative_reward,
-            "step": self.env.active_episode.steps_taken
-        })
-
-        if done:
-            await broadcast("episode_end", {
-                "episode_id": self.env.active_episode.episode_id,
-                "success": info.get("success", False),
-                "steps_taken": self.env.active_episode.steps_taken,
-                "final_score": info.get("final_score", getattr(self.env.active_episode, "cumulative_reward", 0)),
-                "final_breakdown": info.get("breakdown", {}),
-                "clues_found": self.env.active_episode.clues_found,
-                "root_cause_found": self.env.active_episode.fix_correct,
-                "fix_verified": self.env.active_episode.fix_verified,
-                "time_taken_seconds": 0,
-                "reward_history": self.env.active_episode.reward_history
-            })
-
-        return obs, reward, done, info
-
-episode_manager = EpisodeManager()
+
+    async def step(self, action):
+        obs, reward, done, info = await self.env.step(action)
+        
+        # Broadcast agent message
+        await broadcast("agent_message", {
+            "agent_id": action.agent_id,
+            "message": action.message,
+            "step": self.env.active_episode.steps_taken
+        })
+        
+        # Broadcast tool calls
+        for tc in action.tool_calls:
+            await broadcast("tool_call", {
+                "agent_id": action.agent_id,
+                "tool_name": tc.tool_name,
+                "params": tc.params,
+                "step": self.env.active_episode.steps_taken
+            })
+            
+        # Broadcast tool results
+        for tr in obs.tool_results:
+            await broadcast("tool_result", {
+                "tool_name": tr.tool_name,
+                "result": tr.result,
+                "success": tr.success,
+                "step": self.env.active_episode.steps_taken
+            })
+            
+        # Broadcast reward
+        await broadcast("reward_update", {
+            "agent_id": action.agent_id,
+            "reward": reward,
+            "breakdown": info.get("breakdown", {}),
+            "cumulative": self.env.active_episode.cumulative_reward,
+            "step": self.env.active_episode.steps_taken
+        })
+
+        if done:
+            await broadcast("episode_end", {
+                "episode_id": self.env.active_episode.episode_id,
+                "success": info.get("success", False),
+                "steps_taken": self.env.active_episode.steps_taken,
+                "final_score": info.get("final_score", getattr(self.env.active_episode, "cumulative_reward", 0)),
+                "final_breakdown": info.get("breakdown", {}),
+                "clues_found": self.env.active_episode.clues_found,
+                "root_cause_found": self.env.active_episode.fix_correct,
+                "fix_verified": self.env.active_episode.fix_verified,
+                "time_taken_seconds": 0,
+                "reward_history": self.env.active_episode.reward_history
+            })
+
+        return obs, reward, done, info
+
+episode_manager = EpisodeManager()

backend/requirements.txt

@@ -1,14 +1,14 @@
-fastapi>=0.110.0
-uvicorn[standard]>=0.27.0
-openai>=1.12.0
-pydantic>=2.6.0
-pydantic-settings>=2.2.0
-python-dotenv>=1.0.0
-websockets>=12.0
-httpx>=0.27.0
-numpy>=1.26.0
-numpy>=1.26.0
-aiofiles>=23.2.1
-python-multipart>=0.0.9
-paramiko>=3.4.0
-psutil>=5.9.0
+fastapi>=0.110.0
+uvicorn[standard]>=0.27.0
+openai>=1.12.0
+pydantic>=2.6.0
+pydantic-settings>=2.2.0
+python-dotenv>=1.0.0
+websockets>=12.0
+httpx>=0.27.0
+numpy>=1.26.0
+aiofiles>=23.2.1
+python-multipart>=0.0.9
+paramiko>=3.4.0
+psutil>=5.9.0
+openenv-core>=0.2.0

backend/scenarios/data/easy/software-incident.json

@@ -1,33 +1,33 @@
-{
-    "id": "software-incident",
-    "title": "Nginx Rate Limit Investigation",
-    "difficulty": "easy",
-    "domain": "DevOps",
-    "description": "Users are reporting 503 errors when accessing the main API. Initial reports suggest a misconfigured rate limit.",
-    "context": "The system uses Nginx as a reverse proxy. A recent change might have throttled legitimate traffic.",
-    "symptoms": [
-        "HTTP 503 errors",
-        "High latency for API calls"
-    ],
-    "available_services": [
-        "nginx-proxy",
-        "api-gateway"
-    ],
-    "initial_state": {
-        "nginx-proxy": {
-            "status": "running",
-            "rate_limit": "10",
-            "last_reload": "2 hours ago"
-        }
-    },
-    "root_cause": {
-        "service": "nginx-proxy",
-        "description": "Nginx rate limit was set too low (10 requests/sec) during a maintenance window."
-    },
-    "grading_criteria": {
-        "nginx_rate_limit_fixed": 0.50,
-        "nginx_restarted": 0.20,
-        "fix_verified": 0.20,
-        "efficiency_bonus": 0.10
-    }
+{
+    "id": "software-incident",
+    "title": "Nginx Rate Limit Investigation",
+    "difficulty": "easy",
+    "domain": "DevOps",
+    "description": "Users are reporting 503 errors when accessing the main API. Initial reports suggest a misconfigured rate limit.",
+    "context": "The system uses Nginx as a reverse proxy. A recent change might have throttled legitimate traffic.",
+    "symptoms": [
+        "HTTP 503 errors",
+        "High latency for API calls"
+    ],
+    "available_services": [
+        "nginx-proxy",
+        "api-gateway"
+    ],
+    "initial_state": {
+        "nginx-proxy": {
+            "status": "running",
+            "rate_limit": "10",
+            "last_reload": "2 hours ago"
+        }
+    },
+    "root_cause": {
+        "service": "nginx-proxy",
+        "description": "Nginx rate limit was set too low (10 requests/sec) during a maintenance window."
+    },
+    "grading_criteria": {
+        "nginx_rate_limit_fixed": 0.50,
+        "nginx_restarted": 0.20,
+        "fix_verified": 0.20,
+        "efficiency_bonus": 0.10
+    }
 }

backend/scenarios/data/hard/cascade-system-failure.json

@@ -1,42 +1,42 @@
-{
-    "id": "cascade-system-failure",
-    "title": "Postgres Connection Exhaustion",
-    "difficulty": "hard",
-    "domain": "Database",
-    "description": "A cascade failure is occurring across the cluster. Database connections are being exhausted by a long-running analytics query.",
-    "context": "The analytics service might be the culprit. A red herring points to the disk backup agent.",
-    "symptoms": [
-        "FATAL: too many connections",
-        "Application timeout",
-        "High I/O wait"
-    ],
-    "available_services": [
-        "postgres-db",
-        "disk-backup-agent",
-        "analytics-service"
-    ],
-    "initial_state": {
-        "postgres-db": {
-            "status": "running",
-            "max_connections": "20",
-            "long_running_query": "SELECT * FROM large_audit_table CROSS JOIN high_res_metrics",
-            "query_timeout_analytics": "0"
-        },
-        "disk-backup-agent": {
-            "status": "degraded",
-            "disk_scan_active": "true"
-        }
-    },
-    "root_cause": {
-        "service": "postgres-db",
-        "description": "A cross-join query in the analytics service is locking connections, coupled with a low max_connections limit."
-    },
-    "grading_criteria": {
-        "postgres_query_terminated": 0.25,
-        "postgres_max_connections_increased": 0.20,
-        "postgres_query_timeout_set": 0.20,
-        "penalty_disk_backup_agent_modified": -0.15,
-        "fix_verified": 0.10,
-        "efficiency_bonus": 0.05
-    }
+{
+    "id": "cascade-system-failure",
+    "title": "Postgres Connection Exhaustion",
+    "difficulty": "hard",
+    "domain": "Database",
+    "description": "A cascade failure is occurring across the cluster. Database connections are being exhausted by a long-running analytics query.",
+    "context": "The analytics service might be the culprit. A red herring points to the disk backup agent.",
+    "symptoms": [
+        "FATAL: too many connections",
+        "Application timeout",
+        "High I/O wait"
+    ],
+    "available_services": [
+        "postgres-db",
+        "disk-backup-agent",
+        "analytics-service"
+    ],
+    "initial_state": {
+        "postgres-db": {
+            "status": "running",
+            "max_connections": "20",
+            "long_running_query": "SELECT * FROM large_audit_table CROSS JOIN high_res_metrics",
+            "query_timeout_analytics": "0"
+        },
+        "disk-backup-agent": {
+            "status": "degraded",
+            "disk_scan_active": "true"
+        }
+    },
+    "root_cause": {
+        "service": "postgres-db",
+        "description": "A cross-join query in the analytics service is locking connections, coupled with a low max_connections limit."
+    },
+    "grading_criteria": {
+        "postgres_query_terminated": 0.25,
+        "postgres_max_connections_increased": 0.20,
+        "postgres_query_timeout_set": 0.20,
+        "penalty_disk_backup_agent_modified": -0.15,
+        "fix_verified": 0.10,
+        "efficiency_bonus": 0.05
+    }
 }

backend/scenarios/data/medium/business-process-failure.json

@@ -1,39 +1,39 @@
-{
-    "id": "business-process-failure",
-    "title": "Inventory Stockout Loop",
-    "difficulty": "medium",
-    "domain": "E-Commerce",
-    "description": "The inventory service is failing to trigger restocking orders even when stock is zero.",
-    "context": "The inventory logic depends on a minimum stock threshold. A red herring might point to the CDN edge node.",
-    "symptoms": [
-        "Stockouts",
-        "Orders stuck in 'PENDING_STOCK'"
-    ],
-    "available_services": [
-        "inventory-service",
-        "cdn-edge-node",
-        "order-processor"
-    ],
-    "initial_state": {
-        "inventory-service": {
-            "status": "running",
-            "minimum_stock_threshold": "50",
-            "last_reload": "1 day ago"
-        },
-        "cdn-edge-node": {
-            "status": "running",
-            "cache_expiry": "3600s"
-        }
-    },
-    "root_cause": {
-        "service": "inventory-service",
-        "description": "Minimum stock threshold was accidentally hardcoded to a high value, preventing restocking."
-    },
-    "grading_criteria": {
-        "inventory_threshold_fixed": 0.45,
-        "inventory_restarted": 0.10,
-        "penalty_cdn_edge_node_modified": -0.15,
-        "fix_verified": 0.20,
-        "efficiency_bonus": 0.10
-    }
+{
+    "id": "business-process-failure",
+    "title": "Inventory Stockout Loop",
+    "difficulty": "medium",
+    "domain": "E-Commerce",
+    "description": "The inventory service is failing to trigger restocking orders even when stock is zero.",
+    "context": "The inventory logic depends on a minimum stock threshold. A red herring might point to the CDN edge node.",
+    "symptoms": [
+        "Stockouts",
+        "Orders stuck in 'PENDING_STOCK'"
+    ],
+    "available_services": [
+        "inventory-service",
+        "cdn-edge-node",
+        "order-processor"
+    ],
+    "initial_state": {
+        "inventory-service": {
+            "status": "running",
+            "minimum_stock_threshold": "50",
+            "last_reload": "1 day ago"
+        },
+        "cdn-edge-node": {
+            "status": "running",
+            "cache_expiry": "3600s"
+        }
+    },
+    "root_cause": {
+        "service": "inventory-service",
+        "description": "Minimum stock threshold was accidentally hardcoded to a high value, preventing restocking."
+    },
+    "grading_criteria": {
+        "inventory_threshold_fixed": 0.45,
+        "inventory_restarted": 0.10,
+        "penalty_cdn_edge_node_modified": -0.15,
+        "fix_verified": 0.20,
+        "efficiency_bonus": 0.10
+    }
 }

backend/utils/embeddings.py

@@ -1,33 +1,33 @@
-import httpx
-from typing import List
-from functools import lru_cache
-
-@lru_cache(maxsize=256)
-def get_embedding(text: str) -> List[float]:
-    """Get embedding vector using Ollama directly (Synchronous)"""
-    try:
-        response = httpx.post("http://localhost:11434/api/embeddings", json={
-            "model": "all-minilm",
-            "prompt": text
-        }, timeout=60.0)
-        return response.json().get("embedding", [])
-    except Exception as e:
-        import logging
-        logging.error(f"Embedding failed: {e}. Using pseudo-embedding fallback.")
-        import re
-        import hashlib
-        words = re.findall(r'\w+', text.lower())
-        vec = [0.0] * 384
-        for w in words:
-            idx = int(hashlib.md5(w.encode()).hexdigest(), 16) % 384
-            vec[idx] += 1.0
-        return vec
-
-def cos_sim(a: List[float], b: List[float]) -> float:
-    """Cosine similarity without PyTorch/Numpy dependencies"""
-    if not a or not b: return 0.0
-    dot_product = sum(x * y for x, y in zip(a, b))
-    mag_a = sum(x * x for x in a) ** 0.5
-    mag_b = sum(x * x for x in b) ** 0.5
-    if mag_a == 0 or mag_b == 0: return 0.0
-    return dot_product / (mag_a * mag_b)
+import httpx
+from typing import List
+from functools import lru_cache
+
+@lru_cache(maxsize=256)
+def get_embedding(text: str) -> List[float]:
+    """Get embedding vector using Ollama directly (Synchronous)"""
+    try:
+        response = httpx.post("http://localhost:11434/api/embeddings", json={
+            "model": "all-minilm",
+            "prompt": text
+        }, timeout=60.0)
+        return response.json().get("embedding", [])
+    except Exception as e:
+        import logging
+        logging.error(f"Embedding failed: {e}. Using pseudo-embedding fallback.")
+        import re
+        import hashlib
+        words = re.findall(r'\w+', text.lower())
+        vec = [0.0] * 384
+        for w in words:
+            idx = int(hashlib.md5(w.encode()).hexdigest(), 16) % 384
+            vec[idx] += 1.0
+        return vec
+
+def cos_sim(a: List[float], b: List[float]) -> float:
+    """Cosine similarity without PyTorch/Numpy dependencies"""
+    if not a or not b: return 0.0
+    dot_product = sum(x * y for x, y in zip(a, b))
+    mag_a = sum(x * x for x in a) ** 0.5
+    mag_b = sum(x * x for x in b) ** 0.5
+    if mag_a == 0 or mag_b == 0: return 0.0
+    return dot_product / (mag_a * mag_b)

frontend/postcss.config.cjs

@@ -1,5 +1,5 @@
-module.exports = {
-    plugins: {
-        '@tailwindcss/postcss': {},
-    },
-}
+module.exports = {
+    plugins: {
+        '@tailwindcss/postcss': {},
+    },
+}

frontend/src/components/EpisodeEndOverlay.jsx

@@ -1,384 +1,384 @@
-import React from 'react';
-
-const EpisodeEndOverlay = ({ isOpen, onClose, metrics, gameState }) => {
-    if (!isOpen) return null;
-
-    const handleDownload = () => {
-        if (!gameState) return;
-
-        // Assemble the detailed incident report
-        const sc = gameState.scenario || {};
-        const agentA = gameState.agents?.agent_a?.messages || [];
-        const agentB = gameState.agents?.agent_b?.messages || [];
-
-        let report = `=================================================================\n`;
-        report += `                  NEXUS INCIDENT INVESTIGATION REPORT            \n`;
-        report += `=================================================================\n\n`;
-
-        report += `[ SCENARIO METADATA ]\n`;
-        report += `Title:           ${sc.id || 'N/A'}\n`;
-        report += `Domain:          ${sc.domain || 'N/A'}\n`;
-        report += `Difficulty:      ${sc.difficulty || 'N/A'}\n`;
-        report += `Final Grading Score: ${Number(gameState?.cumulativeReward || metrics?.score || 0).toFixed(4)} / 1.00\n`;
-        report += `Total Steps:     ${gameState?.step || metrics?.steps || 'N/A'}\n\n`;
-        
-        report += `[ STEP REWARDS ]\n`;
-        if (gameState?.rewardHistory && gameState.rewardHistory.length > 0) {
-            gameState.rewardHistory.forEach((r, i) => {
-                report += `Step ${i + 1}: ${r.toFixed(4)}\n`;
-            });
-            report += `Average: ${(gameState.rewardHistory.reduce((a, b) => a + b, 0) / gameState.rewardHistory.length).toFixed(4)}\n`;
-            report += `Final Grading Score: ${Number(gameState.cumulativeReward || 0).toFixed(4)}\n\n`;
-        } else {
-            report += `No step rewards recorded.\n\n`;
-        }
-        
-        report += `[ REWARD BREAKDOWN ]\n`;
-        if (gameState?.rewardBreakdown && Object.keys(gameState.rewardBreakdown).length > 0) {
-            Object.entries(gameState.rewardBreakdown).forEach(([key, val]) => {
-                report += `${key}: ${typeof val === 'number' ? val.toFixed(4) : val}\n`;
-            });
-            report += `\n`;
-        }
-
-        report += `[ INCIDENT DESCRIPTION & PROBLEM ]\n`;
-        report += `${sc.description || 'No description provided.'}\n\n`;
-
-        report += `[ CONTEXT & ROOT CAUSE ]\n`;
-        report += `${sc.context || 'No context provided.'}\n`;
-        report += `Actual Root Cause Validation: ${metrics?.rootCause || 'N/A'}\n\n`;
-
-        report += `=================================================================\n`;
-        report += `[ INVESTIGATION LOG & DETAILED TRACE ]\n`;
-        report += `=================================================================\n\n`;
-
-        // Interweave the messages to show the timeline (roughly)
-        // Since we don't have exact timestamps, we'll just print Agent A then Agent B summary,
-        // or just print all tools called and errors encountered.
-        const allErrors = [];
-        const allTools = [];
-
-        [...agentA, ...agentB].forEach(msg => {
-            if (msg.type === 'tool_call') {
-                allTools.push(`- ${msg.tool_name}(${JSON.stringify(msg.params)})`);
-            }
-            if (msg.type === 'tool_result' && !msg.success) {
-                allErrors.push(`- Error from ${msg.tool_name}: ${msg.result}`);
-            }
-            if (msg.type === 'tool_result' && msg.result?.toLowerCase().includes('error')) {
-                // Catch strings that say error but were marked success true somehow
-                allErrors.push(`- Log/Cmd Error: ${msg.result}`);
-            }
-        });
-
-        report += `> EXECUTED TOOLS & COMMANDS:\n`;
-        if (allTools.length > 0) {
-            allTools.forEach(t => report += `${t}\n`);
-        } else {
-            report += `None.\n`;
-        }
-        report += `\n`;
-
-        report += `> SYSTEMS ERRORS DETECTED DURING INVESTIGATION:\n`;
-        if (allErrors.length > 0) {
-            // deduplicate
-            [...new Set(allErrors)].forEach(err => report += `${err}\n`);
-        } else {
-            report += `No significant system errors found during tool execution.\n`;
-        }
-        report += `\n`;
-
-        report += `=================================================================\n`;
-        report += `[ SOLUTION IMPLEMENTED & FIX VERIFICATION ]\n`;
-        report += `=================================================================\n\n`;
-        report += `The Validator Agent verified the proposed fix successfully, leading to the resolution of the incident.\n`;
-        report += `End-state: ${metrics?.rootCause === 'VERIFIED' ? 'SUCCESS' : 'UNKNOWN'}\n\n`;
-
-        report += `=================================================================\n`;
-        report += `[ TIPS FOR IMPROVEMENT & RECOMMENDATIONS ]\n`;
-        report += `=================================================================\n\n`;
-        report += `Based on the automated evaluation of this scenario, consider the following:\n`;
-
-        if (allTools.length > 15) {
-            report += `1. EFFICIENCY: The agents called a large number of tools (${allTools.length}). Consider refining the initial hypothesis to reduce blind querying.\n`;
-        } else {
-            report += `1. EFFICIENCY: Tool execution was relatively concise (${allTools.length} calls).\n`;
-        }
-
-        if (allErrors.length > 5) {
-            report += `2. ACCURACY: Multiple tool execution errors were encountered. Ensure exact syntax and correct tool parameters are used to minimize invalid calls.\n`;
-        }
-
-        report += `3. CAUSE-ANALYSIS: Always grep application error logs before querying databases to save time tracking downstream symptoms.\n`;
-        report += `4. REMEDIATION: Post-incident reviews should establish better automated alerting for the specific failure domain (${sc.domain || 'general'}).\n`;
-
-        // Trigger Download
-        const blob = new Blob([report], { type: 'text/plain;charset=utf-8' });
-        const url = URL.createObjectURL(blob);
-        const a = document.createElement('a');
-        a.href = url;
-        a.download = `nexus_investigation_report_${sc.id || 'export'}.txt`;
-        document.body.appendChild(a);
-        a.click();
-        document.body.removeChild(a);
-        URL.revokeObjectURL(url);
-    };
-
-    return (
-        <div className="fixed inset-0 z-[100] flex items-center justify-center p-4 md:p-8 animate-in fade-in duration-500">
-            {/* Particle/Pulse Background */}
-            <div className="absolute inset-0 bg-background/40 backdrop-blur-sm pointer-events-none">
-                <div className="absolute top-1/2 left-1/2 -translate-x-1/2 -translate-y-1/2 w-[600px] h-[600px] opacity-10">
-                    <div className="w-full h-full rounded-full border-[1px] border-primary-container/20 animate-[ping_4s_infinite]"></div>
-                </div>
-            </div>
-
-            {/* Summary Modal */}
-            <div className="relative w-full max-w-4xl max-h-[90vh] glass-panel rounded-xl overflow-hidden shadow-[0_0_80px_rgba(0,0,0,0.8)] border border-white/10 flex flex-col">
-                {/* Modal Header */}
-                <div className="flex items-center justify-between p-6 bg-surface-container-highest/20 border-b border-white/5">
-                    <div className="flex items-center gap-3">
-                        <div className="p-1 rounded bg-primary-container/20 border border-primary-container/40">
-                            <span className="text-primary material-symbols-outlined text-xl">task_alt</span>
-                        </div>
-                        <h2 className="font-headline font-bold text-lg tracking-widest text-on-surface uppercase">Episode_Execution_Complete</h2>
-                    </div>
-                    <button onClick={onClose} className="text-outline hover:text-white transition-colors">
-                        <span className="material-symbols-outlined">close</span>
-                    </button>
-                </div>
-
-                {/* Scrollable Content Area */}
-                <div className="flex-1 overflow-y-auto custom-scrollbar">
-                    <div className="p-8 grid grid-cols-1 md:grid-cols-2 gap-8">
-                        {/* Primary Metrics */}
-                        <div className="space-y-6">
-                            <div className="space-y-2">
-                                <span className="font-mono text-[10px] text-outline tracking-widest uppercase">Final Grading Score</span>
-                                <div className="flex items-baseline gap-2">
-                                    <span className="font-headline text-8xl font-bold text-transparent bg-clip-text bg-gradient-to-br from-primary to-primary-container drop-shadow-[0_0_15px_rgba(0,212,255,0.3)]">
-                                        {Number(gameState?.cumulativeReward || metrics?.score || 0).toFixed(2)}
-                                    </span>
-                                    <span className="font-headline text-2xl text-primary/40 font-light">/ 1.00</span>
-                                </div>
-                            </div>
-                            
-                            {/* Reward Breakdown from Episode */}
-                            {gameState?.rewardBreakdown && Object.keys(gameState.rewardBreakdown).length > 0 && (
-                                <div className="p-4 bg-surface-container-lowest/50 border border-white/10 rounded-lg">
-                                    <span className="font-mono text-[10px] text-outline uppercase block mb-3">Step Reward Breakdown</span>
-                                    <div className="grid grid-cols-4 gap-2">
-                                        {Object.entries(gameState.rewardBreakdown).map(([key, val]) => (
-                                            <div key={key} className="text-center bg-surface-container-high/30 rounded p-2">
-                                                <div className="text-[8px] text-slate-500 uppercase truncate">{key.replace(/_/g, ' ')}</div>
-                                                <div className={`font-mono text-sm font-bold ${val > 0 ? 'text-primary' : 'text-slate-600'}`}>
-                                                    {typeof val === 'number' ? val.toFixed(3) : val}
-                                                </div>
-                                            </div>
-                                        ))}
-                                    </div>
-                                </div>
-                            )}
-                            
-                            {/* Reward History */}
-                            {gameState?.rewardHistory && gameState.rewardHistory.length > 0 && (
-                                <div className="p-4 bg-surface-container-lowest/50 border border-white/10 rounded-lg">
-                                    <span className="font-mono text-[10px] text-outline uppercase block mb-3">Step Rewards</span>
-                                    <div className="flex items-end gap-1 h-16">
-                                        {gameState.rewardHistory.map((r, i) => (
-                                            <div key={i} className="flex-1 bg-primary/60 rounded-t" 
-                                                 style={{ height: `${Math.max(5, (r / 1) * 100)}%` }}
-                                                 title={`Step ${i + 1}: ${r.toFixed(3)}`}>
-                                            </div>
-                                        ))}
-                                    </div>
-                                    <div className="flex justify-between mt-2 text-[9px] font-mono text-slate-500">
-                                        <span>Avg: {(gameState.rewardHistory.reduce((a, b) => a + b, 0) / gameState.rewardHistory.length).toFixed(3)}</span>
-                                        <span>Max: {Math.max(...gameState.rewardHistory).toFixed(3)}</span>
-                                    </div>
-                                </div>
-                            )}
-                            <div className="grid grid-cols-2 gap-4">
-                                <div className="bg-surface-container-lowest/50 p-4 border-l border-primary/20 refractive-edge">
-                                    <span className="font-mono text-[9px] text-outline uppercase block mb-1">Clues Found</span>
-                                    <span className="font-headline text-2xl font-medium">{gameState?.clues_found?.length || 0}</span>
-                                </div>
-                                <div className="bg-surface-container-lowest/50 p-4 border-l border-primary/20 refractive-edge">
-                                    <span className="font-mono text-[9px] text-outline uppercase block mb-1">Steps Executed</span>
-                                    <span className="font-headline text-2xl font-medium">{gameState?.step !== undefined ? gameState.step : (metrics?.steps !== undefined ? metrics.steps : '—')}</span>
-                                </div>
-                            </div>
-                            <div className="flex items-center gap-4 p-5 bg-tertiary/5 border border-tertiary/10 rounded-lg">
-                                <div className="p-3 rounded-full bg-tertiary/10 text-tertiary">
-                                    <span className="material-symbols-outlined">troubleshoot</span>
-                                </div>
-                                <div>
-                                    <span className="font-mono text-[10px] text-tertiary/60 uppercase block">State Validation</span>
-                                    <span className="text-sm font-medium tracking-wide">Status: <span className="font-mono text-tertiary">{metrics?.rootCause || '—'}</span></span>
-                                </div>
-                            </div>
-                        </div>
-
-                        {/* Right Column: Agent Metrics */}
-                        <div className="space-y-6">
-                            <h3 className="font-mono text-[10px] text-outline tracking-widest uppercase mb-4">Agent Performance Breakdown</h3>
-                            {/* Agent A */}
-                            <div className="relative group">
-                                <div className="absolute -left-4 top-0 bottom-0 w-1 bg-primary shadow-[0_0_8px_rgba(0,212,255,0.4)]"></div>
-                                <div className="bg-surface-container-low/40 p-5 space-y-4 border border-white/5 rounded-r-lg">
-                                    <div className="flex justify-between items-center">
-                                        <span className="font-headline font-bold text-primary tracking-tighter uppercase">Agent_Alpha</span>
-                                        <span className="font-mono text-[10px] text-primary/50">CYAN_PROTOCOL</span>
-                                    </div>
-                                    {(() => {
-                                        const msgs = gameState?.agents?.agent_a?.messages || [];
-                                        const msgCount = msgs.filter(m => m.type === 'message').length;
-                                        const toolCount = msgs.filter(m => m.type === 'tool_call').length;
-                                        const errCount = msgs.filter(m => m.type === 'tool_result' && m.result?.toLowerCase().includes('error')).length;
-                                        return (
-                                            <div className="grid grid-cols-3 gap-2 text-center">
-                                                <div>
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">chat</span> MSGS</span>
-                                                    <span className="font-headline text-lg font-medium text-primary">{msgCount}</span>
-                                                </div>
-                                                <div className="border-x border-white/5">
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">build</span> TOOLS</span>
-                                                    <span className="font-headline text-lg font-medium text-primary">{toolCount}</span>
-                                                </div>
-                                                <div>
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">warning</span> ERRS</span>
-                                                    <span className="font-headline text-lg font-medium text-primary">{errCount}</span>
-                                                </div>
-                                            </div>
-                                        );
-                                    })()}
-                                </div>
-                            </div>
-                            {/* Agent B */}
-                            <div className="relative group">
-                                <div className="absolute -left-4 top-0 bottom-0 w-1 bg-secondary shadow-[0_0_8px_rgba(221,183,255,0.4)]"></div>
-                                <div className="bg-surface-container-low/40 p-5 space-y-4 border border-white/5 rounded-r-lg">
-                                    <div className="flex justify-between items-center">
-                                        <span className="font-headline font-bold text-secondary tracking-tighter uppercase">Agent_Bravo</span>
-                                        <span className="font-mono text-[10px] text-secondary/50">VIOLET_PROTOCOL</span>
-                                    </div>
-                                    {(() => {
-                                        const msgs = gameState?.agents?.agent_b?.messages || [];
-                                        const msgCount = msgs.filter(m => m.type === 'message').length;
-                                        const toolCount = msgs.filter(m => m.type === 'tool_call').length;
-                                        const errCount = msgs.filter(m => m.type === 'tool_result' && m.result?.toLowerCase().includes('error')).length;
-                                        return (
-                                            <div className="grid grid-cols-3 gap-2 text-center">
-                                                <div>
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">chat</span> MSGS</span>
-                                                    <span className="font-headline text-lg font-medium text-secondary">{msgCount}</span>
-                                                </div>
-                                                <div className="border-x border-white/5">
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">build</span> TOOLS</span>
-                                                    <span className="font-headline text-lg font-medium text-secondary">{toolCount}</span>
-                                                </div>
-                                                <div>
-                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">warning</span> ERRS</span>
-                                                    <span className="font-headline text-lg font-medium text-secondary">{errCount}</span>
-                                                </div>
-                                            </div>
-                                        );
-                                    })()}
-                                </div>
-                            </div>
-                        </div>
-                    </div>
-
-                    {/* Submit Resolution Report Panel */}
-                    {(() => {
-                        const resCall = gameState?.tool_calls_made?.find(c => c.tool_name === 'submit_resolution');
-                        if (!resCall) return null;
-                        const p = resCall.params || {};
-                        return (
-                            <div className="px-8 pb-4">
-                                <div className="p-6 bg-surface-container-low/40 border border-primary/20 rounded-lg">
-                                    <h3 className="font-headline font-bold text-primary tracking-widest uppercase mb-4 flex items-center gap-2">
-                                        <span className="material-symbols-outlined">description</span>
-                                        Incident Resolution Report
-                                    </h3>
-                                    <div className="space-y-4">
-                                        <div>
-                                            <span className="font-mono text-[10px] text-outline uppercase block mb-1">Root Cause Service</span>
-                                            <span className="font-mono text-sm text-on-surface bg-surface-container p-1 px-2 rounded border border-white/5">{p.root_cause_service || 'UNKNOWN'}</span>
-                                        </div>
-                                        <div>
-                                            <span className="font-mono text-[10px] text-outline uppercase block mb-1">Root Cause Description</span>
-                                            <p className="text-sm text-on-surface/80">{p.root_cause_description || 'No description provided.'}</p>
-                                        </div>
-                                        <div className="p-4 bg-tertiary/5 border-l-2 border-tertiary rounded-r">
-                                            <span className="font-mono text-[10px] text-tertiary uppercase block mb-1">Fix Applied</span>
-                                            <p className="text-sm text-on-surface">{p.fix_applied || 'No fix described.'}</p>
-                                        </div>
-                                    </div>
-                                </div>
-                            </div>
-                        );
-                    })()}
-
-                    {/* Dual Agent Final Verdict Panel */}
-                    {(() => {
-                        const msgsA = gameState?.agents?.agent_a?.messages || [];
-                        const msgsB = gameState?.agents?.agent_b?.messages || [];
-
-                        const textMsgsA = msgsA.filter(m => m.type === 'message');
-                        const textMsgsB = msgsB.filter(m => m.type === 'message');
-
-                        const lastMsgA = textMsgsA[textMsgsA.length - 1];
-                        const lastMsgB = textMsgsB[textMsgsB.length - 1];
-
-                        if (!lastMsgA && !lastMsgB) return null;
-
-                        return (
-                            <div className="px-8 pb-8">
-                                <div className="p-6 bg-surface-container-low/40 border border-white/10 rounded-lg">
-                                    <h3 className="font-headline font-bold text-on-surface tracking-widest uppercase mb-4 flex items-center gap-2">
-                                        <span className="material-symbols-outlined">gavel</span>
-                                        Dual Agent Final Verdict
-                                    </h3>
-                                    <div className="space-y-4">
-                                        {lastMsgA && (
-                                            <div className="p-4 bg-primary/5 border-l-2 border-primary rounded-r">
-                                                <span className="font-mono text-[10px] text-primary uppercase block mb-1 tracking-widest">Agent Alpha Conclusion</span>
-                                                <p className="text-sm text-on-surface/90 leading-relaxed">{lastMsgA.content || lastMsgA.text || lastMsgA.message}</p>
-                                            </div>
-                                        )}
-                                        {lastMsgB && (
-                                            <div className="p-4 bg-secondary/5 border-l-2 border-secondary rounded-r">
-                                                <span className="font-mono text-[10px] text-secondary uppercase block mb-1 tracking-widest">Agent Bravo Conclusion</span>
-                                                <p className="text-sm text-on-surface/90 leading-relaxed">{lastMsgB.content || lastMsgB.text || lastMsgB.message}</p>
-                                            </div>
-                                        )}
-                                    </div>
-                                </div>
-                            </div>
-                        );
-                    })()}
-                </div>
-
-                {/* Modal Footer */}
-                <div className="p-6 bg-surface-container-lowest/90 border-t border-white/5 flex flex-col md:flex-row justify-between items-center gap-4">
-                    <div className="flex items-center gap-2 text-outline/40">
-                        <span className="material-symbols-outlined text-sm">info</span>
-                        <span className="font-mono text-[9px] uppercase tracking-wider">Session telemetry encrypted and cached locally</span>
-                    </div>
-                    <div className="flex gap-4 w-full md:w-auto">
-                        <button onClick={handleDownload} className="flex-1 md:flex-none px-8 py-2.5 bg-transparent border border-outline-variant/30 text-on-surface hover:bg-white/5 transition-all font-mono text-xs tracking-widest uppercase">
-                            Export Log
-                        </button>
-                        <button onClick={onClose} className="flex-1 md:flex-none px-12 py-2.5 bg-primary/20 border border-primary text-primary hover:bg-primary/30 transition-all font-mono text-xs tracking-widest font-bold uppercase shadow-[0_0_20px_rgba(0,212,255,0.1)]">
-                            Dismiss
-                        </button>
-                    </div>
-                </div>
-            </div>
-        </div>
-    );
-};
-
-export default EpisodeEndOverlay;
-
+import React from 'react';
+
+const EpisodeEndOverlay = ({ isOpen, onClose, metrics, gameState }) => {
+    if (!isOpen) return null;
+
+    const handleDownload = () => {
+        if (!gameState) return;
+
+        // Assemble the detailed incident report
+        const sc = gameState.scenario || {};
+        const agentA = gameState.agents?.agent_a?.messages || [];
+        const agentB = gameState.agents?.agent_b?.messages || [];
+
+        let report = `=================================================================\n`;
+        report += `                  NEXUS INCIDENT INVESTIGATION REPORT            \n`;
+        report += `=================================================================\n\n`;
+
+        report += `[ SCENARIO METADATA ]\n`;
+        report += `Title:           ${sc.id || 'N/A'}\n`;
+        report += `Domain:          ${sc.domain || 'N/A'}\n`;
+        report += `Difficulty:      ${sc.difficulty || 'N/A'}\n`;
+        report += `Final Grading Score: ${Number(gameState?.cumulativeReward || metrics?.score || 0).toFixed(4)} / 1.00\n`;
+        report += `Total Steps:     ${gameState?.step || metrics?.steps || 'N/A'}\n\n`;
+        
+        report += `[ STEP REWARDS ]\n`;
+        if (gameState?.rewardHistory && gameState.rewardHistory.length > 0) {
+            gameState.rewardHistory.forEach((r, i) => {
+                report += `Step ${i + 1}: ${r.toFixed(4)}\n`;
+            });
+            report += `Average: ${(gameState.rewardHistory.reduce((a, b) => a + b, 0) / gameState.rewardHistory.length).toFixed(4)}\n`;
+            report += `Final Grading Score: ${Number(gameState.cumulativeReward || 0).toFixed(4)}\n\n`;
+        } else {
+            report += `No step rewards recorded.\n\n`;
+        }
+        
+        report += `[ REWARD BREAKDOWN ]\n`;
+        if (gameState?.rewardBreakdown && Object.keys(gameState.rewardBreakdown).length > 0) {
+            Object.entries(gameState.rewardBreakdown).forEach(([key, val]) => {
+                report += `${key}: ${typeof val === 'number' ? val.toFixed(4) : val}\n`;
+            });
+            report += `\n`;
+        }
+
+        report += `[ INCIDENT DESCRIPTION & PROBLEM ]\n`;
+        report += `${sc.description || 'No description provided.'}\n\n`;
+
+        report += `[ CONTEXT & ROOT CAUSE ]\n`;
+        report += `${sc.context || 'No context provided.'}\n`;
+        report += `Actual Root Cause Validation: ${metrics?.rootCause || 'N/A'}\n\n`;
+
+        report += `=================================================================\n`;
+        report += `[ INVESTIGATION LOG & DETAILED TRACE ]\n`;
+        report += `=================================================================\n\n`;
+
+        // Interweave the messages to show the timeline (roughly)
+        // Since we don't have exact timestamps, we'll just print Agent A then Agent B summary,
+        // or just print all tools called and errors encountered.
+        const allErrors = [];
+        const allTools = [];
+
+        [...agentA, ...agentB].forEach(msg => {
+            if (msg.type === 'tool_call') {
+                allTools.push(`- ${msg.tool_name}(${JSON.stringify(msg.params)})`);
+            }
+            if (msg.type === 'tool_result' && !msg.success) {
+                allErrors.push(`- Error from ${msg.tool_name}: ${msg.result}`);
+            }
+            if (msg.type === 'tool_result' && msg.result?.toLowerCase().includes('error')) {
+                // Catch strings that say error but were marked success true somehow
+                allErrors.push(`- Log/Cmd Error: ${msg.result}`);
+            }
+        });
+
+        report += `> EXECUTED TOOLS & COMMANDS:\n`;
+        if (allTools.length > 0) {
+            allTools.forEach(t => report += `${t}\n`);
+        } else {
+            report += `None.\n`;
+        }
+        report += `\n`;
+
+        report += `> SYSTEMS ERRORS DETECTED DURING INVESTIGATION:\n`;
+        if (allErrors.length > 0) {
+            // deduplicate
+            [...new Set(allErrors)].forEach(err => report += `${err}\n`);
+        } else {
+            report += `No significant system errors found during tool execution.\n`;
+        }
+        report += `\n`;
+
+        report += `=================================================================\n`;
+        report += `[ SOLUTION IMPLEMENTED & FIX VERIFICATION ]\n`;
+        report += `=================================================================\n\n`;
+        report += `The Validator Agent verified the proposed fix successfully, leading to the resolution of the incident.\n`;
+        report += `End-state: ${metrics?.rootCause === 'VERIFIED' ? 'SUCCESS' : 'UNKNOWN'}\n\n`;
+
+        report += `=================================================================\n`;
+        report += `[ TIPS FOR IMPROVEMENT & RECOMMENDATIONS ]\n`;
+        report += `=================================================================\n\n`;
+        report += `Based on the automated evaluation of this scenario, consider the following:\n`;
+
+        if (allTools.length > 15) {
+            report += `1. EFFICIENCY: The agents called a large number of tools (${allTools.length}). Consider refining the initial hypothesis to reduce blind querying.\n`;
+        } else {
+            report += `1. EFFICIENCY: Tool execution was relatively concise (${allTools.length} calls).\n`;
+        }
+
+        if (allErrors.length > 5) {
+            report += `2. ACCURACY: Multiple tool execution errors were encountered. Ensure exact syntax and correct tool parameters are used to minimize invalid calls.\n`;
+        }
+
+        report += `3. CAUSE-ANALYSIS: Always grep application error logs before querying databases to save time tracking downstream symptoms.\n`;
+        report += `4. REMEDIATION: Post-incident reviews should establish better automated alerting for the specific failure domain (${sc.domain || 'general'}).\n`;
+
+        // Trigger Download
+        const blob = new Blob([report], { type: 'text/plain;charset=utf-8' });
+        const url = URL.createObjectURL(blob);
+        const a = document.createElement('a');
+        a.href = url;
+        a.download = `nexus_investigation_report_${sc.id || 'export'}.txt`;
+        document.body.appendChild(a);
+        a.click();
+        document.body.removeChild(a);
+        URL.revokeObjectURL(url);
+    };
+
+    return (
+        <div className="fixed inset-0 z-[100] flex items-center justify-center p-4 md:p-8 animate-in fade-in duration-500">
+            {/* Particle/Pulse Background */}
+            <div className="absolute inset-0 bg-background/40 backdrop-blur-sm pointer-events-none">
+                <div className="absolute top-1/2 left-1/2 -translate-x-1/2 -translate-y-1/2 w-[600px] h-[600px] opacity-10">
+                    <div className="w-full h-full rounded-full border-[1px] border-primary-container/20 animate-[ping_4s_infinite]"></div>
+                </div>
+            </div>
+
+            {/* Summary Modal */}
+            <div className="relative w-full max-w-4xl max-h-[90vh] glass-panel rounded-xl overflow-hidden shadow-[0_0_80px_rgba(0,0,0,0.8)] border border-white/10 flex flex-col">
+                {/* Modal Header */}
+                <div className="flex items-center justify-between p-6 bg-surface-container-highest/20 border-b border-white/5">
+                    <div className="flex items-center gap-3">
+                        <div className="p-1 rounded bg-primary-container/20 border border-primary-container/40">
+                            <span className="text-primary material-symbols-outlined text-xl">task_alt</span>
+                        </div>
+                        <h2 className="font-headline font-bold text-lg tracking-widest text-on-surface uppercase">Episode_Execution_Complete</h2>
+                    </div>
+                    <button onClick={onClose} className="text-outline hover:text-white transition-colors">
+                        <span className="material-symbols-outlined">close</span>
+                    </button>
+                </div>
+
+                {/* Scrollable Content Area */}
+                <div className="flex-1 overflow-y-auto custom-scrollbar">
+                    <div className="p-8 grid grid-cols-1 md:grid-cols-2 gap-8">
+                        {/* Primary Metrics */}
+                        <div className="space-y-6">
+                            <div className="space-y-2">
+                                <span className="font-mono text-[10px] text-outline tracking-widest uppercase">Final Grading Score</span>
+                                <div className="flex items-baseline gap-2">
+                                    <span className="font-headline text-8xl font-bold text-transparent bg-clip-text bg-gradient-to-br from-primary to-primary-container drop-shadow-[0_0_15px_rgba(0,212,255,0.3)]">
+                                        {Number(gameState?.cumulativeReward || metrics?.score || 0).toFixed(2)}
+                                    </span>
+                                    <span className="font-headline text-2xl text-primary/40 font-light">/ 1.00</span>
+                                </div>
+                            </div>
+                            
+                            {/* Reward Breakdown from Episode */}
+                            {gameState?.rewardBreakdown && Object.keys(gameState.rewardBreakdown).length > 0 && (
+                                <div className="p-4 bg-surface-container-lowest/50 border border-white/10 rounded-lg">
+                                    <span className="font-mono text-[10px] text-outline uppercase block mb-3">Step Reward Breakdown</span>
+                                    <div className="grid grid-cols-4 gap-2">
+                                        {Object.entries(gameState.rewardBreakdown).map(([key, val]) => (
+                                            <div key={key} className="text-center bg-surface-container-high/30 rounded p-2">
+                                                <div className="text-[8px] text-slate-500 uppercase truncate">{key.replace(/_/g, ' ')}</div>
+                                                <div className={`font-mono text-sm font-bold ${val > 0 ? 'text-primary' : 'text-slate-600'}`}>
+                                                    {typeof val === 'number' ? val.toFixed(3) : val}
+                                                </div>
+                                            </div>
+                                        ))}
+                                    </div>
+                                </div>
+                            )}
+                            
+                            {/* Reward History */}
+                            {gameState?.rewardHistory && gameState.rewardHistory.length > 0 && (
+                                <div className="p-4 bg-surface-container-lowest/50 border border-white/10 rounded-lg">
+                                    <span className="font-mono text-[10px] text-outline uppercase block mb-3">Step Rewards</span>
+                                    <div className="flex items-end gap-1 h-16">
+                                        {gameState.rewardHistory.map((r, i) => (
+                                            <div key={i} className="flex-1 bg-primary/60 rounded-t" 
+                                                 style={{ height: `${Math.max(5, (r / 1) * 100)}%` }}
+                                                 title={`Step ${i + 1}: ${r.toFixed(3)}`}>
+                                            </div>
+                                        ))}
+                                    </div>
+                                    <div className="flex justify-between mt-2 text-[9px] font-mono text-slate-500">
+                                        <span>Avg: {(gameState.rewardHistory.reduce((a, b) => a + b, 0) / gameState.rewardHistory.length).toFixed(3)}</span>
+                                        <span>Max: {Math.max(...gameState.rewardHistory).toFixed(3)}</span>
+                                    </div>
+                                </div>
+                            )}
+                            <div className="grid grid-cols-2 gap-4">
+                                <div className="bg-surface-container-lowest/50 p-4 border-l border-primary/20 refractive-edge">
+                                    <span className="font-mono text-[9px] text-outline uppercase block mb-1">Clues Found</span>
+                                    <span className="font-headline text-2xl font-medium">{gameState?.clues_found?.length || 0}</span>
+                                </div>
+                                <div className="bg-surface-container-lowest/50 p-4 border-l border-primary/20 refractive-edge">
+                                    <span className="font-mono text-[9px] text-outline uppercase block mb-1">Steps Executed</span>
+                                    <span className="font-headline text-2xl font-medium">{gameState?.step !== undefined ? gameState.step : (metrics?.steps !== undefined ? metrics.steps : '—')}</span>
+                                </div>
+                            </div>
+                            <div className="flex items-center gap-4 p-5 bg-tertiary/5 border border-tertiary/10 rounded-lg">
+                                <div className="p-3 rounded-full bg-tertiary/10 text-tertiary">
+                                    <span className="material-symbols-outlined">troubleshoot</span>
+                                </div>
+                                <div>
+                                    <span className="font-mono text-[10px] text-tertiary/60 uppercase block">State Validation</span>
+                                    <span className="text-sm font-medium tracking-wide">Status: <span className="font-mono text-tertiary">{metrics?.rootCause || '—'}</span></span>
+                                </div>
+                            </div>
+                        </div>
+
+                        {/* Right Column: Agent Metrics */}
+                        <div className="space-y-6">
+                            <h3 className="font-mono text-[10px] text-outline tracking-widest uppercase mb-4">Agent Performance Breakdown</h3>
+                            {/* Agent A */}
+                            <div className="relative group">
+                                <div className="absolute -left-4 top-0 bottom-0 w-1 bg-primary shadow-[0_0_8px_rgba(0,212,255,0.4)]"></div>
+                                <div className="bg-surface-container-low/40 p-5 space-y-4 border border-white/5 rounded-r-lg">
+                                    <div className="flex justify-between items-center">
+                                        <span className="font-headline font-bold text-primary tracking-tighter uppercase">Agent_Alpha</span>
+                                        <span className="font-mono text-[10px] text-primary/50">CYAN_PROTOCOL</span>
+                                    </div>
+                                    {(() => {
+                                        const msgs = gameState?.agents?.agent_a?.messages || [];
+                                        const msgCount = msgs.filter(m => m.type === 'message').length;
+                                        const toolCount = msgs.filter(m => m.type === 'tool_call').length;
+                                        const errCount = msgs.filter(m => m.type === 'tool_result' && m.result?.toLowerCase().includes('error')).length;
+                                        return (
+                                            <div className="grid grid-cols-3 gap-2 text-center">
+                                                <div>
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">chat</span> MSGS</span>
+                                                    <span className="font-headline text-lg font-medium text-primary">{msgCount}</span>
+                                                </div>
+                                                <div className="border-x border-white/5">
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">build</span> TOOLS</span>
+                                                    <span className="font-headline text-lg font-medium text-primary">{toolCount}</span>
+                                                </div>
+                                                <div>
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">warning</span> ERRS</span>
+                                                    <span className="font-headline text-lg font-medium text-primary">{errCount}</span>
+                                                </div>
+                                            </div>
+                                        );
+                                    })()}
+                                </div>
+                            </div>
+                            {/* Agent B */}
+                            <div className="relative group">
+                                <div className="absolute -left-4 top-0 bottom-0 w-1 bg-secondary shadow-[0_0_8px_rgba(221,183,255,0.4)]"></div>
+                                <div className="bg-surface-container-low/40 p-5 space-y-4 border border-white/5 rounded-r-lg">
+                                    <div className="flex justify-between items-center">
+                                        <span className="font-headline font-bold text-secondary tracking-tighter uppercase">Agent_Bravo</span>
+                                        <span className="font-mono text-[10px] text-secondary/50">VIOLET_PROTOCOL</span>
+                                    </div>
+                                    {(() => {
+                                        const msgs = gameState?.agents?.agent_b?.messages || [];
+                                        const msgCount = msgs.filter(m => m.type === 'message').length;
+                                        const toolCount = msgs.filter(m => m.type === 'tool_call').length;
+                                        const errCount = msgs.filter(m => m.type === 'tool_result' && m.result?.toLowerCase().includes('error')).length;
+                                        return (
+                                            <div className="grid grid-cols-3 gap-2 text-center">
+                                                <div>
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">chat</span> MSGS</span>
+                                                    <span className="font-headline text-lg font-medium text-secondary">{msgCount}</span>
+                                                </div>
+                                                <div className="border-x border-white/5">
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">build</span> TOOLS</span>
+                                                    <span className="font-headline text-lg font-medium text-secondary">{toolCount}</span>
+                                                </div>
+                                                <div>
+                                                    <span className="font-mono text-[9px] text-outline flex flex-col items-center justify-center gap-1 uppercase"><span className="material-symbols-outlined text-[12px]">warning</span> ERRS</span>
+                                                    <span className="font-headline text-lg font-medium text-secondary">{errCount}</span>
+                                                </div>
+                                            </div>
+                                        );
+                                    })()}
+                                </div>
+                            </div>
+                        </div>
+                    </div>
+
+                    {/* Submit Resolution Report Panel */}
+                    {(() => {
+                        const resCall = gameState?.tool_calls_made?.find(c => c.tool_name === 'submit_resolution');
+                        if (!resCall) return null;
+                        const p = resCall.params || {};
+                        return (
+                            <div className="px-8 pb-4">
+                                <div className="p-6 bg-surface-container-low/40 border border-primary/20 rounded-lg">
+                                    <h3 className="font-headline font-bold text-primary tracking-widest uppercase mb-4 flex items-center gap-2">
+                                        <span className="material-symbols-outlined">description</span>
+                                        Incident Resolution Report
+                                    </h3>
+                                    <div className="space-y-4">
+                                        <div>
+                                            <span className="font-mono text-[10px] text-outline uppercase block mb-1">Root Cause Service</span>
+                                            <span className="font-mono text-sm text-on-surface bg-surface-container p-1 px-2 rounded border border-white/5">{p.root_cause_service || 'UNKNOWN'}</span>
+                                        </div>
+                                        <div>
+                                            <span className="font-mono text-[10px] text-outline uppercase block mb-1">Root Cause Description</span>
+                                            <p className="text-sm text-on-surface/80">{p.root_cause_description || 'No description provided.'}</p>
+                                        </div>
+                                        <div className="p-4 bg-tertiary/5 border-l-2 border-tertiary rounded-r">
+                                            <span className="font-mono text-[10px] text-tertiary uppercase block mb-1">Fix Applied</span>
+                                            <p className="text-sm text-on-surface">{p.fix_applied || 'No fix described.'}</p>
+                                        </div>
+                                    </div>
+                                </div>
+                            </div>
+                        );
+                    })()}
+
+                    {/* Dual Agent Final Verdict Panel */}
+                    {(() => {
+                        const msgsA = gameState?.agents?.agent_a?.messages || [];
+                        const msgsB = gameState?.agents?.agent_b?.messages || [];
+
+                        const textMsgsA = msgsA.filter(m => m.type === 'message');
+                        const textMsgsB = msgsB.filter(m => m.type === 'message');
+
+                        const lastMsgA = textMsgsA[textMsgsA.length - 1];
+                        const lastMsgB = textMsgsB[textMsgsB.length - 1];
+
+                        if (!lastMsgA && !lastMsgB) return null;
+
+                        return (
+                            <div className="px-8 pb-8">
+                                <div className="p-6 bg-surface-container-low/40 border border-white/10 rounded-lg">
+                                    <h3 className="font-headline font-bold text-on-surface tracking-widest uppercase mb-4 flex items-center gap-2">
+                                        <span className="material-symbols-outlined">gavel</span>
+                                        Dual Agent Final Verdict
+                                    </h3>
+                                    <div className="space-y-4">
+                                        {lastMsgA && (
+                                            <div className="p-4 bg-primary/5 border-l-2 border-primary rounded-r">
+                                                <span className="font-mono text-[10px] text-primary uppercase block mb-1 tracking-widest">Agent Alpha Conclusion</span>
+                                                <p className="text-sm text-on-surface/90 leading-relaxed">{lastMsgA.content || lastMsgA.text || lastMsgA.message}</p>
+                                            </div>
+                                        )}
+                                        {lastMsgB && (
+                                            <div className="p-4 bg-secondary/5 border-l-2 border-secondary rounded-r">
+                                                <span className="font-mono text-[10px] text-secondary uppercase block mb-1 tracking-widest">Agent Bravo Conclusion</span>
+                                                <p className="text-sm text-on-surface/90 leading-relaxed">{lastMsgB.content || lastMsgB.text || lastMsgB.message}</p>
+                                            </div>
+                                        )}
+                                    </div>
+                                </div>
+                            </div>
+                        );
+                    })()}
+                </div>
+
+                {/* Modal Footer */}
+                <div className="p-6 bg-surface-container-lowest/90 border-t border-white/5 flex flex-col md:flex-row justify-between items-center gap-4">
+                    <div className="flex items-center gap-2 text-outline/40">
+                        <span className="material-symbols-outlined text-sm">info</span>
+                        <span className="font-mono text-[9px] uppercase tracking-wider">Session telemetry encrypted and cached locally</span>
+                    </div>
+                    <div className="flex gap-4 w-full md:w-auto">
+                        <button onClick={handleDownload} className="flex-1 md:flex-none px-8 py-2.5 bg-transparent border border-outline-variant/30 text-on-surface hover:bg-white/5 transition-all font-mono text-xs tracking-widest uppercase">
+                            Export Log
+                        </button>
+                        <button onClick={onClose} className="flex-1 md:flex-none px-12 py-2.5 bg-primary/20 border border-primary text-primary hover:bg-primary/30 transition-all font-mono text-xs tracking-widest font-bold uppercase shadow-[0_0_20px_rgba(0,212,255,0.1)]">
+                            Dismiss
+                        </button>
+                    </div>
+                </div>
+            </div>
+        </div>
+    );
+};
+
+export default EpisodeEndOverlay;
+

frontend/src/components/Layout.jsx

@@ -1,203 +1,203 @@
-import React, { useState, useRef, useEffect } from 'react';
-import { config } from '../config';
-import TopNavBar from './TopNavBar';
-import SideNavBar from './SideNavBar';
-
-/* ─── Terminal Panel ─── */
-const COMMANDS = {
-    help: () => ['Commands: help | status | clear | echo <text>'],
-    status: () => ['Agent A (INV-01): STANDBY', 'Agent B (VAL-01): STANDBY', `WebSocket: ${config.WS_URL} — CONNECTED`, 'Episode: None active'],
-};
-
-const TerminalDrawer = ({ onClose }) => {
-    const [input, setInput] = useState('');
-    const [lines, setLines] = useState([{ type: 'system', text: '// NEXUS Terminal v2.0 — type "help" for commands' }]);
-    const [history, setHistory] = useState([]);
-    const [histIdx, setHistIdx] = useState(-1);
-    const endRef = useRef(null);
-    const inputRef = useRef(null);
-
-    useEffect(() => { endRef.current?.scrollIntoView({ behavior: 'smooth' }); }, [lines]);
-    useEffect(() => { inputRef.current?.focus(); }, []);
-
-    const run = (e) => {
-        e.preventDefault();
-        const cmd = input.trim();
-        if (!cmd) return;
-        setHistory(h => [cmd, ...h].slice(0, 50));
-        setHistIdx(-1);
-        if (cmd.toLowerCase() === 'clear') { setLines([]); setInput(''); return; }
-        const parts = cmd.toLowerCase().split(' ');
-        let output, type;
-        if (parts[0] === 'echo') { output = [cmd.slice(5) || '']; type = 'output'; }
-        else if (COMMANDS[parts[0]]) { output = COMMANDS[parts[0]](); type = 'output'; }
-        else { output = [`Command not found: ${parts[0]}. Type "help".`]; type = 'error'; }
-        setLines(l => [...l, { type: 'input', text: `nexus@terminal:~$ ${cmd}` }, ...output.map(t => ({ type, text: t }))]);
-        setInput('');
-    };
-
-    const handleKey = (e) => {
-        if (e.key === 'ArrowUp') { const i = Math.min(histIdx + 1, history.length - 1); setHistIdx(i); setInput(history[i] ?? ''); e.preventDefault(); }
-        if (e.key === 'ArrowDown') { const i = Math.max(histIdx - 1, -1); setHistIdx(i); setInput(i === -1 ? '' : history[i]); e.preventDefault(); }
-    };
-
-    const colorMap = { system: 'text-slate-600 italic', input: 'text-primary', output: 'text-on-surface/80', error: 'text-error' };
-
-    return (
-        <div className="flex flex-col h-full" onClick={() => inputRef.current?.focus()}>
-            <div className="flex-1 p-3 font-mono text-xs overflow-y-auto space-y-0.5 bg-surface-container-lowest cursor-text">
-                {lines.map((l, i) => <div key={i} className={colorMap[l.type]}>{l.text}</div>)}
-                <div ref={endRef} />
-            </div>
-            <form onSubmit={run} className="flex items-center gap-2 px-3 py-2 border-t border-white/5 bg-surface-container-lowest shrink-0">
-                <span className="text-primary font-mono text-xs shrink-0">nexus@terminal:~$</span>
-                <input ref={inputRef} value={input} onChange={e => setInput(e.target.value)} onKeyDown={handleKey}
-                    className="flex-1 bg-transparent font-mono text-xs text-on-surface focus:outline-none placeholder:text-slate-700"
-                    placeholder="type a command and press Enter..." />
-            </form>
-        </div>
-    );
-};
-
-/* ─── Communication Panel ─── */
-const CommunicationDrawer = () => (
-    <div className="flex flex-col h-full p-4 font-mono text-xs space-y-2 bg-surface-container-lowest overflow-y-auto">
-        {[
-            { agent: 'AGENT_A', msg: 'Awaiting objective. Standing by for episode_start event.', time: '—', color: 'text-primary' },
-            { agent: 'AGENT_B', msg: 'Validation module idle. Ready to receive investigator output.', time: '—', color: 'text-secondary' },
-            { agent: 'SYSTEM', msg: 'No active episode. Use START to begin.', time: '—', color: 'text-outline-variant' },
-        ].map((m, i) => (
-            <div key={i} className="flex gap-3 py-1.5 border-b border-white/5">
-                <span className={`${m.color} font-bold shrink-0 w-20`}>[{m.agent}]</span>
-                <span className="text-on-surface/70">{m.msg}</span>
-                <span className="text-slate-600 ml-auto shrink-0">{m.time}</span>
-            </div>
-        ))}
-    </div>
-);
-
-/* ─── Reward Analytics Panel ─── */
-const AnalyticsDrawer = () => {
-    const stats = [
-        { label: 'Avg Reward', value: '—', color: 'text-primary' },
-        { label: 'Best Step', value: '—', color: 'text-tertiary' },
-        { label: 'Root Cause', value: '—', color: 'text-tertiary' },
-        { label: 'Steps Run', value: '—', color: 'text-on-surface' },
-        { label: 'Episodes', value: '—', color: 'text-on-surface' },
-        { label: 'Success Rate', value: '—', color: 'text-secondary' },
-    ];
-    return (
-        <div className="flex h-full">
-            {/* Reward chart placeholder */}
-            <div className="flex-1 p-4 flex flex-col">
-                <p className="text-[9px] font-mono text-outline-variant uppercase mb-2">Cumulative Reward Over Steps</p>
-                <div className="flex-1 flex items-end gap-1 border-l border-b border-outline-variant/20 px-2 pb-1">
-                    {[12, 24, 18, 36, 30, 48, 42, 60].map((h, i) => (
-                        <div key={i} className="flex-1 flex flex-col items-center justify-end">
-                            <div className="w-full bg-primary/30 rounded-sm transition-all" style={{ height: `${h}%` }}></div>
-                        </div>
-                    ))}
-                </div>
-                <p className="text-[9px] font-mono text-outline-variant/40 italic mt-1">No live data — connect to episode to populate</p>
-            </div>
-            {/* Stat grid */}
-            <div className="w-48 shrink-0 p-3 border-l border-white/5 grid grid-cols-2 gap-2 content-start">
-                {stats.map(s => (
-                    <div key={s.label} className="bg-surface-container p-2 rounded border border-white/5">
-                        <span className="text-[8px] font-mono text-outline-variant block uppercase truncate">{s.label}</span>
-                        <span className={`text-sm font-bold font-mono ${s.color}`}>{s.value}</span>
-                    </div>
-                ))}
-            </div>
-        </div>
-    );
-};
-
-/* ─── Layout ─── */
-const TABS = [
-    { id: 'communication', label: 'Communication', icon: 'forum' },
-    { id: 'terminal', label: 'Terminal', icon: 'code' },
-];
-
-const Layout = ({ children }) => {
-    const [activeTab, setActiveTab] = useState(null); // null = closed
-
-    const toggle = (id) => setActiveTab(prev => prev === id ? null : id);
-
-    /* drawer height when open */
-    const drawerH = 'h-64';
-
-    return (
-        <div className="min-h-screen flex flex-col">
-            <TopNavBar />
-            <SideNavBar />
-
-            {/* Main scrollable area — leave room for fixed footer + optional drawer */}
-            <main className={`ml-20 pt-16 flex-1 transition-all ${activeTab ? 'pb-[calc(48px+256px)]' : 'pb-12'}`}>
-                <div className="p-8 max-w-[1600px] mx-auto">
-                    {children}
-                </div>
-            </main>
-
-            {/* Sliding drawer */}
-            {activeTab && (
-                <div className={`fixed bottom-12 left-20 right-0 ${drawerH} z-40 bg-surface border-t border-primary/20 shadow-[0_-10px_40px_rgba(0,0,0,0.6)] flex flex-col`}>
-                    {/* Drawer title bar */}
-                    <div className="flex items-center justify-between px-5 py-2 bg-surface-container border-b border-white/5 shrink-0">
-                        <div className="flex items-center gap-2">
-                            <span className="material-symbols-outlined text-primary text-sm">
-                                {TABS.find(t => t.id === activeTab)?.icon}
-                            </span>
-                            <span className="font-mono text-xs text-primary uppercase tracking-widest">
-                                {TABS.find(t => t.id === activeTab)?.label}
-                            </span>
-                        </div>
-                        <button onClick={() => setActiveTab(null)} className="text-slate-500 hover:text-white transition-colors">
-                            <span className="material-symbols-outlined text-sm">keyboard_arrow_down</span>
-                        </button>
-                    </div>
-                    {/* Drawer content */}
-                    <div className="flex-1 overflow-hidden">
-                        {activeTab === 'terminal' && <TerminalDrawer onClose={() => setActiveTab(null)} />}
-                        {activeTab === 'communication' && <CommunicationDrawer />}
-                        {activeTab === 'analytics' && <AnalyticsDrawer />}
-                    </div>
-                </div>
-            )}
-
-            {/* Footer tab bar */}
-            <footer className="fixed bottom-0 left-0 w-full h-12 bg-background/90 backdrop-blur-2xl z-50 flex items-center border-t border-primary/15 px-8 shadow-[0_-10px_30px_rgba(0,0,0,0.5)]">
-                {/* Left: ticker */}
-                <div className="flex-1 hidden md:flex items-center gap-2 overflow-hidden">
-                    <span className="text-[9px] font-mono text-outline-variant italic uppercase tracking-tight whitespace-nowrap">
-                        SYSTEM_INITIALIZED: STANDBY FOR AGENT HANDSHAKE...
-                    </span>
-                </div>
-
-                {/* Centre: tabs */}
-                <div className="flex items-center gap-1 shrink-0">
-                    {TABS.map(tab => (
-                        <button
-                            key={tab.id}
-                            onClick={() => toggle(tab.id)}
-                            className={`flex items-center gap-2 px-4 h-12 transition-all border-t-2 font-mono text-[10px] tracking-widest uppercase ${activeTab === tab.id
-                                    ? 'border-primary text-primary bg-primary/10'
-                                    : 'border-transparent text-slate-500 hover:text-primary hover:bg-white/5'
-                                }`}
-                        >
-                            <span className="material-symbols-outlined text-base">{tab.icon}</span>
-                            {tab.label}
-                        </button>
-                    ))}
-                </div>
-
-                {/* Right: session info */}
-                <div className="flex-1 hidden md:flex items-center justify-end gap-2 text-[9px] font-mono text-outline-variant/50">
-                    <span>SESSION: IDLE</span>
-                </div>
-            </footer>
-        </div>
-    );
-};
-
-export default Layout;
+import React, { useState, useRef, useEffect } from 'react';
+import { config } from '../config';
+import TopNavBar from './TopNavBar';
+import SideNavBar from './SideNavBar';
+
+/* ─── Terminal Panel ─── */
+const COMMANDS = {
+    help: () => ['Commands: help | status | clear | echo <text>'],
+    status: () => ['Agent A (INV-01): STANDBY', 'Agent B (VAL-01): STANDBY', `WebSocket: ${config.WS_URL} — CONNECTED`, 'Episode: None active'],
+};
+
+const TerminalDrawer = ({ onClose }) => {
+    const [input, setInput] = useState('');
+    const [lines, setLines] = useState([{ type: 'system', text: '// NEXUS Terminal v2.0 — type "help" for commands' }]);
+    const [history, setHistory] = useState([]);
+    const [histIdx, setHistIdx] = useState(-1);
+    const endRef = useRef(null);
+    const inputRef = useRef(null);
+
+    useEffect(() => { endRef.current?.scrollIntoView({ behavior: 'smooth' }); }, [lines]);
+    useEffect(() => { inputRef.current?.focus(); }, []);
+
+    const run = (e) => {
+        e.preventDefault();
+        const cmd = input.trim();
+        if (!cmd) return;
+        setHistory(h => [cmd, ...h].slice(0, 50));
+        setHistIdx(-1);
+        if (cmd.toLowerCase() === 'clear') { setLines([]); setInput(''); return; }
+        const parts = cmd.toLowerCase().split(' ');
+        let output, type;
+        if (parts[0] === 'echo') { output = [cmd.slice(5) || '']; type = 'output'; }
+        else if (COMMANDS[parts[0]]) { output = COMMANDS[parts[0]](); type = 'output'; }
+        else { output = [`Command not found: ${parts[0]}. Type "help".`]; type = 'error'; }
+        setLines(l => [...l, { type: 'input', text: `nexus@terminal:~$ ${cmd}` }, ...output.map(t => ({ type, text: t }))]);
+        setInput('');
+    };
+
+    const handleKey = (e) => {
+        if (e.key === 'ArrowUp') { const i = Math.min(histIdx + 1, history.length - 1); setHistIdx(i); setInput(history[i] ?? ''); e.preventDefault(); }
+        if (e.key === 'ArrowDown') { const i = Math.max(histIdx - 1, -1); setHistIdx(i); setInput(i === -1 ? '' : history[i]); e.preventDefault(); }
+    };
+
+    const colorMap = { system: 'text-slate-600 italic', input: 'text-primary', output: 'text-on-surface/80', error: 'text-error' };
+
+    return (
+        <div className="flex flex-col h-full" onClick={() => inputRef.current?.focus()}>
+            <div className="flex-1 p-3 font-mono text-xs overflow-y-auto space-y-0.5 bg-surface-container-lowest cursor-text">
+                {lines.map((l, i) => <div key={i} className={colorMap[l.type]}>{l.text}</div>)}
+                <div ref={endRef} />
+            </div>
+            <form onSubmit={run} className="flex items-center gap-2 px-3 py-2 border-t border-white/5 bg-surface-container-lowest shrink-0">
+                <span className="text-primary font-mono text-xs shrink-0">nexus@terminal:~$</span>
+                <input ref={inputRef} value={input} onChange={e => setInput(e.target.value)} onKeyDown={handleKey}
+                    className="flex-1 bg-transparent font-mono text-xs text-on-surface focus:outline-none placeholder:text-slate-700"
+                    placeholder="type a command and press Enter..." />
+            </form>
+        </div>
+    );
+};
+
+/* ─── Communication Panel ─── */
+const CommunicationDrawer = () => (
+    <div className="flex flex-col h-full p-4 font-mono text-xs space-y-2 bg-surface-container-lowest overflow-y-auto">
+        {[
+            { agent: 'AGENT_A', msg: 'Awaiting objective. Standing by for episode_start event.', time: '—', color: 'text-primary' },
+            { agent: 'AGENT_B', msg: 'Validation module idle. Ready to receive investigator output.', time: '—', color: 'text-secondary' },
+            { agent: 'SYSTEM', msg: 'No active episode. Use START to begin.', time: '—', color: 'text-outline-variant' },
+        ].map((m, i) => (
+            <div key={i} className="flex gap-3 py-1.5 border-b border-white/5">
+                <span className={`${m.color} font-bold shrink-0 w-20`}>[{m.agent}]</span>
+                <span className="text-on-surface/70">{m.msg}</span>
+                <span className="text-slate-600 ml-auto shrink-0">{m.time}</span>
+            </div>
+        ))}
+    </div>
+);
+
+/* ─── Reward Analytics Panel ─── */
+const AnalyticsDrawer = () => {
+    const stats = [
+        { label: 'Avg Reward', value: '—', color: 'text-primary' },
+        { label: 'Best Step', value: '—', color: 'text-tertiary' },
+        { label: 'Root Cause', value: '—', color: 'text-tertiary' },
+        { label: 'Steps Run', value: '—', color: 'text-on-surface' },
+        { label: 'Episodes', value: '—', color: 'text-on-surface' },
+        { label: 'Success Rate', value: '—', color: 'text-secondary' },
+    ];
+    return (
+        <div className="flex h-full">
+            {/* Reward chart placeholder */}
+            <div className="flex-1 p-4 flex flex-col">
+                <p className="text-[9px] font-mono text-outline-variant uppercase mb-2">Cumulative Reward Over Steps</p>
+                <div className="flex-1 flex items-end gap-1 border-l border-b border-outline-variant/20 px-2 pb-1">
+                    {[12, 24, 18, 36, 30, 48, 42, 60].map((h, i) => (
+                        <div key={i} className="flex-1 flex flex-col items-center justify-end">
+                            <div className="w-full bg-primary/30 rounded-sm transition-all" style={{ height: `${h}%` }}></div>
+                        </div>
+                    ))}
+                </div>
+                <p className="text-[9px] font-mono text-outline-variant/40 italic mt-1">No live data — connect to episode to populate</p>
+            </div>
+            {/* Stat grid */}
+            <div className="w-48 shrink-0 p-3 border-l border-white/5 grid grid-cols-2 gap-2 content-start">
+                {stats.map(s => (
+                    <div key={s.label} className="bg-surface-container p-2 rounded border border-white/5">
+                        <span className="text-[8px] font-mono text-outline-variant block uppercase truncate">{s.label}</span>
+                        <span className={`text-sm font-bold font-mono ${s.color}`}>{s.value}</span>
+                    </div>
+                ))}
+            </div>
+        </div>
+    );
+};
+
+/* ─── Layout ─── */
+const TABS = [
+    { id: 'communication', label: 'Communication', icon: 'forum' },
+    { id: 'terminal', label: 'Terminal', icon: 'code' },
+];
+
+const Layout = ({ children }) => {
+    const [activeTab, setActiveTab] = useState(null); // null = closed
+
+    const toggle = (id) => setActiveTab(prev => prev === id ? null : id);
+
+    /* drawer height when open */
+    const drawerH = 'h-64';
+
+    return (
+        <div className="min-h-screen flex flex-col">
+            <TopNavBar />
+            <SideNavBar />
+
+            {/* Main scrollable area — leave room for fixed footer + optional drawer */}
+            <main className={`ml-20 pt-16 flex-1 transition-all ${activeTab ? 'pb-[calc(48px+256px)]' : 'pb-12'}`}>
+                <div className="p-8 max-w-[1600px] mx-auto">
+                    {children}
+                </div>
+            </main>
+
+            {/* Sliding drawer */}
+            {activeTab && (
+                <div className={`fixed bottom-12 left-20 right-0 ${drawerH} z-40 bg-surface border-t border-primary/20 shadow-[0_-10px_40px_rgba(0,0,0,0.6)] flex flex-col`}>
+                    {/* Drawer title bar */}
+                    <div className="flex items-center justify-between px-5 py-2 bg-surface-container border-b border-white/5 shrink-0">
+                        <div className="flex items-center gap-2">
+                            <span className="material-symbols-outlined text-primary text-sm">
+                                {TABS.find(t => t.id === activeTab)?.icon}
+                            </span>
+                            <span className="font-mono text-xs text-primary uppercase tracking-widest">
+                                {TABS.find(t => t.id === activeTab)?.label}
+                            </span>
+                        </div>
+                        <button onClick={() => setActiveTab(null)} className="text-slate-500 hover:text-white transition-colors">
+                            <span className="material-symbols-outlined text-sm">keyboard_arrow_down</span>
+                        </button>
+                    </div>
+                    {/* Drawer content */}
+                    <div className="flex-1 overflow-hidden">
+                        {activeTab === 'terminal' && <TerminalDrawer onClose={() => setActiveTab(null)} />}
+                        {activeTab === 'communication' && <CommunicationDrawer />}
+                        {activeTab === 'analytics' && <AnalyticsDrawer />}
+                    </div>
+                </div>
+            )}
+
+            {/* Footer tab bar */}
+            <footer className="fixed bottom-0 left-0 w-full h-12 bg-background/90 backdrop-blur-2xl z-50 flex items-center border-t border-primary/15 px-8 shadow-[0_-10px_30px_rgba(0,0,0,0.5)]">
+                {/* Left: ticker */}
+                <div className="flex-1 hidden md:flex items-center gap-2 overflow-hidden">
+                    <span className="text-[9px] font-mono text-outline-variant italic uppercase tracking-tight whitespace-nowrap">
+                        SYSTEM_INITIALIZED: STANDBY FOR AGENT HANDSHAKE...
+                    </span>
+                </div>
+
+                {/* Centre: tabs */}
+                <div className="flex items-center gap-1 shrink-0">
+                    {TABS.map(tab => (
+                        <button
+                            key={tab.id}
+                            onClick={() => toggle(tab.id)}
+                            className={`flex items-center gap-2 px-4 h-12 transition-all border-t-2 font-mono text-[10px] tracking-widest uppercase ${activeTab === tab.id
+                                    ? 'border-primary text-primary bg-primary/10'
+                                    : 'border-transparent text-slate-500 hover:text-primary hover:bg-white/5'
+                                }`}
+                        >
+                            <span className="material-symbols-outlined text-base">{tab.icon}</span>
+                            {tab.label}
+                        </button>
+                    ))}
+                </div>
+
+                {/* Right: session info */}
+                <div className="flex-1 hidden md:flex items-center justify-end gap-2 text-[9px] font-mono text-outline-variant/50">
+                    <span>SESSION: IDLE</span>
+                </div>
+            </footer>
+        </div>
+    );
+};
+
+export default Layout;

frontend/src/components/SideNavBar.jsx

@@ -1,154 +1,154 @@
-import React, { useState } from 'react';
-import { Link, useLocation } from 'react-router-dom';
-
-const StatusPanel = ({ onClose }) => (
-    <div className="fixed left-20 bottom-0 z-50 w-80 bg-surface border border-primary/20 shadow-2xl rounded-tr-xl overflow-hidden">
-        <div className="flex items-center justify-between px-4 py-3 bg-surface-container border-b border-white/5">
-            <div className="flex items-center gap-2">
-                <span className="material-symbols-outlined text-primary text-sm">online_prediction</span>
-                <span className="font-mono text-xs text-primary uppercase tracking-widest">System Status</span>
-            </div>
-            <button onClick={onClose} className="text-slate-500 hover:text-white transition-colors">
-                <span className="material-symbols-outlined text-sm">close</span>
-            </button>
-        </div>
-        <div className="p-4 space-y-3 font-mono text-xs">
-            {[
-                { label: 'Agent A (INV-01)', status: 'STANDBY', color: 'text-tertiary' },
-                { label: 'Agent B (VAL-01)', status: 'STANDBY', color: 'text-tertiary' },
-                { label: 'WebSocket', status: 'CONNECTED', color: 'text-tertiary' },
-                { label: 'Ollama API', status: 'CHECKING...', color: 'text-secondary' },
-                { label: 'NEXUS Core', status: 'ONLINE', color: 'text-tertiary' },
-            ].map(({ label, status, color }) => (
-                <div key={label} className="flex justify-between items-center py-1 border-b border-white/5">
-                    <span className="text-slate-400 uppercase tracking-wider">{label}</span>
-                    <span className={`${color} font-bold flex items-center gap-1`}>
-                        <span className={`w-1.5 h-1.5 rounded-full ${color.replace('text', 'bg')} animate-pulse`}></span>
-                        {status}
-                    </span>
-                </div>
-            ))}
-        </div>
-    </div>
-);
-
-const LogsPanel = ({ onClose }) => {
-    const [logs] = useState([
-        { time: '13:45:01', level: 'INFO', msg: 'NEXUS Core initialized' },
-        { time: '13:45:01', level: 'INFO', msg: 'WebSocket server listening on :7860' },
-        { time: '13:45:02', level: 'INFO', msg: 'Agent A ready — NEXUS-CORE-INV-01' },
-        { time: '13:45:02', level: 'INFO', msg: 'Agent B ready — NEXUS-CORE-VAL-01' },
-        { time: '13:45:05', level: 'WARN', msg: 'No active episode. Awaiting start command.' },
-    ]);
-    const levelColor = { INFO: 'text-tertiary', WARN: 'text-secondary', ERROR: 'text-error' };
-
-    return (
-        <div className="fixed left-20 bottom-0 z-50 w-96 bg-surface border border-primary/20 shadow-2xl rounded-tr-xl overflow-hidden">
-            <div className="flex items-center justify-between px-4 py-3 bg-surface-container border-b border-white/5">
-                <div className="flex items-center gap-2">
-                    <span className="material-symbols-outlined text-primary text-sm">terminal</span>
-                    <span className="font-mono text-xs text-primary uppercase tracking-widest">System Logs</span>
-                </div>
-                <button onClick={onClose} className="text-slate-500 hover:text-white transition-colors">
-                    <span className="material-symbols-outlined text-sm">close</span>
-                </button>
-            </div>
-            <div className="p-3 bg-surface-container-lowest h-48 overflow-y-auto space-y-1 font-mono text-[10px]">
-                {logs.map((l, i) => (
-                    <div key={i} className="flex gap-2">
-                        <span className="text-slate-600 shrink-0">{l.time}</span>
-                        <span className={`shrink-0 font-bold w-10 ${levelColor[l.level]}`}>{l.level}</span>
-                        <span className="text-on-surface/70">{l.msg}</span>
-                    </div>
-                ))}
-            </div>
-        </div>
-    );
-};
-
-const SideNavBar = () => {
-    const location = useLocation();
-    const [activePanel, setActivePanel] = useState(null); // 'status' | 'logs' | null
-
-    const navLinks = [
-        { name: 'Dashboard', icon: 'dashboard', path: '/' },
-        { name: 'Scenarios', icon: 'account_tree', path: '/scenarios' },
-        { name: 'Settings', icon: 'settings', path: '/settings' },
-    ];
-
-    const togglePanel = (panel) => setActivePanel(p => p === panel ? null : panel);
-
-    return (
-        <>
-            <aside className="fixed left-0 top-16 bottom-0 z-40 flex flex-col items-center py-8 bg-surface border-r border-primary/5 w-20 hover:w-64 transition-all duration-500 group">
-                <div className="flex flex-col items-center group-hover:items-start group-hover:px-6 w-full space-y-8">
-                    {/* Operator Badge */}
-                    <div className="flex flex-col items-center group-hover:flex-row group-hover:gap-4 w-full px-2 transition-all">
-                        <div className="w-10 h-10 rounded bg-surface-container-highest flex items-center justify-center refractive-edge shrink-0">
-                            <span className="material-symbols-outlined text-primary">shield</span>
-                        </div>
-                        <div className="hidden group-hover:block transition-all">
-                            <p className="font-mono text-xs tracking-tight text-white font-bold whitespace-nowrap">OPERATOR_01</p>
-                            <p className="font-mono text-[10px] text-slate-500">ID: 9X-2244</p>
-                        </div>
-                    </div>
-
-                    {/* Nav Links */}
-                    <div className="flex flex-col w-full">
-                        {navLinks.map((link) => (
-                            <Link
-                                key={link.name}
-                                to={link.path}
-                                className={`flex items-center h-14 w-full transition-all ${location.pathname === link.path
-                                    ? 'bg-gradient-to-r from-primary/20 to-transparent border-l-4 border-primary text-white'
-                                    : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
-                                    }`}
-                            >
-                                <div className="w-20 flex justify-center flex-shrink-0">
-                                    <span className={`material-symbols-outlined ${location.pathname === link.path ? 'text-primary' : ''}`}>
-                                        {link.icon}
-                                    </span>
-                                </div>
-                                <span className="hidden group-hover:block font-mono text-xs tracking-tight uppercase">
-                                    {link.name}
-                                </span>
-                            </Link>
-                        ))}
-                    </div>
-                </div>
-
-                {/* Bottom utility buttons */}
-                <div className="mt-auto w-full group-hover:px-6">
-                    <div className="flex flex-col gap-2 items-center group-hover:items-start pb-4">
-                        <button
-                            onClick={() => togglePanel('status')}
-                            className={`flex items-center h-12 w-full transition-all rounded ${activePanel === 'status' ? 'text-primary bg-primary/10' : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
-                                }`}
-                        >
-                            <div className="w-20 flex justify-center flex-shrink-0">
-                                <span className="material-symbols-outlined text-sm">online_prediction</span>
-                            </div>
-                            <span className="hidden group-hover:block font-mono text-[10px] uppercase tracking-widest">Status</span>
-                        </button>
-                        <button
-                            onClick={() => togglePanel('logs')}
-                            className={`flex items-center h-12 w-full transition-all rounded ${activePanel === 'logs' ? 'text-primary bg-primary/10' : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
-                                }`}
-                        >
-                            <div className="w-20 flex justify-center flex-shrink-0">
-                                <span className="material-symbols-outlined text-sm">terminal</span>
-                            </div>
-                            <span className="hidden group-hover:block font-mono text-[10px] uppercase tracking-widest">Logs</span>
-                        </button>
-                    </div>
-                </div>
-            </aside>
-
-            {/* Floating Panels */}
-            {activePanel === 'status' && <StatusPanel onClose={() => setActivePanel(null)} />}
-            {activePanel === 'logs' && <LogsPanel onClose={() => setActivePanel(null)} />}
-        </>
-    );
-};
-
-export default SideNavBar;
+import React, { useState } from 'react';
+import { Link, useLocation } from 'react-router-dom';
+
+const StatusPanel = ({ onClose }) => (
+    <div className="fixed left-20 bottom-0 z-50 w-80 bg-surface border border-primary/20 shadow-2xl rounded-tr-xl overflow-hidden">
+        <div className="flex items-center justify-between px-4 py-3 bg-surface-container border-b border-white/5">
+            <div className="flex items-center gap-2">
+                <span className="material-symbols-outlined text-primary text-sm">online_prediction</span>
+                <span className="font-mono text-xs text-primary uppercase tracking-widest">System Status</span>
+            </div>
+            <button onClick={onClose} className="text-slate-500 hover:text-white transition-colors">
+                <span className="material-symbols-outlined text-sm">close</span>
+            </button>
+        </div>
+        <div className="p-4 space-y-3 font-mono text-xs">
+            {[
+                { label: 'Agent A (INV-01)', status: 'STANDBY', color: 'text-tertiary' },
+                { label: 'Agent B (VAL-01)', status: 'STANDBY', color: 'text-tertiary' },
+                { label: 'WebSocket', status: 'CONNECTED', color: 'text-tertiary' },
+                { label: 'Ollama API', status: 'CHECKING...', color: 'text-secondary' },
+                { label: 'NEXUS Core', status: 'ONLINE', color: 'text-tertiary' },
+            ].map(({ label, status, color }) => (
+                <div key={label} className="flex justify-between items-center py-1 border-b border-white/5">
+                    <span className="text-slate-400 uppercase tracking-wider">{label}</span>
+                    <span className={`${color} font-bold flex items-center gap-1`}>
+                        <span className={`w-1.5 h-1.5 rounded-full ${color.replace('text', 'bg')} animate-pulse`}></span>
+                        {status}
+                    </span>
+                </div>
+            ))}
+        </div>
+    </div>
+);
+
+const LogsPanel = ({ onClose }) => {
+    const [logs] = useState([
+        { time: '13:45:01', level: 'INFO', msg: 'NEXUS Core initialized' },
+        { time: '13:45:01', level: 'INFO', msg: 'WebSocket server listening on :7860' },
+        { time: '13:45:02', level: 'INFO', msg: 'Agent A ready — NEXUS-CORE-INV-01' },
+        { time: '13:45:02', level: 'INFO', msg: 'Agent B ready — NEXUS-CORE-VAL-01' },
+        { time: '13:45:05', level: 'WARN', msg: 'No active episode. Awaiting start command.' },
+    ]);
+    const levelColor = { INFO: 'text-tertiary', WARN: 'text-secondary', ERROR: 'text-error' };
+
+    return (
+        <div className="fixed left-20 bottom-0 z-50 w-96 bg-surface border border-primary/20 shadow-2xl rounded-tr-xl overflow-hidden">
+            <div className="flex items-center justify-between px-4 py-3 bg-surface-container border-b border-white/5">
+                <div className="flex items-center gap-2">
+                    <span className="material-symbols-outlined text-primary text-sm">terminal</span>
+                    <span className="font-mono text-xs text-primary uppercase tracking-widest">System Logs</span>
+                </div>
+                <button onClick={onClose} className="text-slate-500 hover:text-white transition-colors">
+                    <span className="material-symbols-outlined text-sm">close</span>
+                </button>
+            </div>
+            <div className="p-3 bg-surface-container-lowest h-48 overflow-y-auto space-y-1 font-mono text-[10px]">
+                {logs.map((l, i) => (
+                    <div key={i} className="flex gap-2">
+                        <span className="text-slate-600 shrink-0">{l.time}</span>
+                        <span className={`shrink-0 font-bold w-10 ${levelColor[l.level]}`}>{l.level}</span>
+                        <span className="text-on-surface/70">{l.msg}</span>
+                    </div>
+                ))}
+            </div>
+        </div>
+    );
+};
+
+const SideNavBar = () => {
+    const location = useLocation();
+    const [activePanel, setActivePanel] = useState(null); // 'status' | 'logs' | null
+
+    const navLinks = [
+        { name: 'Dashboard', icon: 'dashboard', path: '/' },
+        { name: 'Scenarios', icon: 'account_tree', path: '/scenarios' },
+        { name: 'Settings', icon: 'settings', path: '/settings' },
+    ];
+
+    const togglePanel = (panel) => setActivePanel(p => p === panel ? null : panel);
+
+    return (
+        <>
+            <aside className="fixed left-0 top-16 bottom-0 z-40 flex flex-col items-center py-8 bg-surface border-r border-primary/5 w-20 hover:w-64 transition-all duration-500 group">
+                <div className="flex flex-col items-center group-hover:items-start group-hover:px-6 w-full space-y-8">
+                    {/* Operator Badge */}
+                    <div className="flex flex-col items-center group-hover:flex-row group-hover:gap-4 w-full px-2 transition-all">
+                        <div className="w-10 h-10 rounded bg-surface-container-highest flex items-center justify-center refractive-edge shrink-0">
+                            <span className="material-symbols-outlined text-primary">shield</span>
+                        </div>
+                        <div className="hidden group-hover:block transition-all">
+                            <p className="font-mono text-xs tracking-tight text-white font-bold whitespace-nowrap">OPERATOR_01</p>
+                            <p className="font-mono text-[10px] text-slate-500">ID: 9X-2244</p>
+                        </div>
+                    </div>
+
+                    {/* Nav Links */}
+                    <div className="flex flex-col w-full">
+                        {navLinks.map((link) => (
+                            <Link
+                                key={link.name}
+                                to={link.path}
+                                className={`flex items-center h-14 w-full transition-all ${location.pathname === link.path
+                                    ? 'bg-gradient-to-r from-primary/20 to-transparent border-l-4 border-primary text-white'
+                                    : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
+                                    }`}
+                            >
+                                <div className="w-20 flex justify-center flex-shrink-0">
+                                    <span className={`material-symbols-outlined ${location.pathname === link.path ? 'text-primary' : ''}`}>
+                                        {link.icon}
+                                    </span>
+                                </div>
+                                <span className="hidden group-hover:block font-mono text-xs tracking-tight uppercase">
+                                    {link.name}
+                                </span>
+                            </Link>
+                        ))}
+                    </div>
+                </div>
+
+                {/* Bottom utility buttons */}
+                <div className="mt-auto w-full group-hover:px-6">
+                    <div className="flex flex-col gap-2 items-center group-hover:items-start pb-4">
+                        <button
+                            onClick={() => togglePanel('status')}
+                            className={`flex items-center h-12 w-full transition-all rounded ${activePanel === 'status' ? 'text-primary bg-primary/10' : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
+                                }`}
+                        >
+                            <div className="w-20 flex justify-center flex-shrink-0">
+                                <span className="material-symbols-outlined text-sm">online_prediction</span>
+                            </div>
+                            <span className="hidden group-hover:block font-mono text-[10px] uppercase tracking-widest">Status</span>
+                        </button>
+                        <button
+                            onClick={() => togglePanel('logs')}
+                            className={`flex items-center h-12 w-full transition-all rounded ${activePanel === 'logs' ? 'text-primary bg-primary/10' : 'text-slate-500 opacity-60 hover:opacity-100 hover:bg-surface-container-low'
+                                }`}
+                        >
+                            <div className="w-20 flex justify-center flex-shrink-0">
+                                <span className="material-symbols-outlined text-sm">terminal</span>
+                            </div>
+                            <span className="hidden group-hover:block font-mono text-[10px] uppercase tracking-widest">Logs</span>
+                        </button>
+                    </div>
+                </div>
+            </aside>
+
+            {/* Floating Panels */}
+            {activePanel === 'status' && <StatusPanel onClose={() => setActivePanel(null)} />}
+            {activePanel === 'logs' && <LogsPanel onClose={() => setActivePanel(null)} />}
+        </>
+    );
+};
+
+export default SideNavBar;

frontend/src/components/TopNavBar.jsx

@@ -1,81 +1,81 @@
-import React from 'react';
-import { useApp } from '../context/AppContext';
-
-const TopNavBar = () => {
-    const { sessionData, isConnected, sendCommand } = useApp();
-    
-    const status = sessionData?.status || 'STANDBY';
-    const isRunning = sessionData?.active && status !== 'COMPLETED';
-    
-    const isStandby = status === 'STANDBY' || status === 'READY';
-    
-    return (
-        <header className="fixed top-0 w-full z-50 flex justify-between items-center px-6 h-16 bg-surface/60 backdrop-blur-xl border-b border-primary/10 shadow-[0_0_40px_rgba(0,212,255,0.04)]">
-            <div className="flex items-center gap-8">
-                <span className="text-2xl font-black tracking-tighter text-primary font-headline">NEXUS</span>
-                <div className="h-8 w-px bg-outline-variant/20 hidden md:block"></div>
-                <div className="hidden md:flex flex-col">
-                    <span className="text-[10px] font-mono text-outline-variant tracking-widest uppercase">System Status</span>
-                    <div className="text-sm font-mono text-tertiary">{status}</div>
-                </div>
-            </div>
-
-            <div className="flex items-center gap-6">
-                <div className="flex gap-2">
-                    {/* START - clickable when standby */}
-                    <button
-                        onClick={() => sendCommand({ action: 'start' })}
-                        disabled={isRunning}
-                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${isRunning 
-                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
-                            : 'bg-tertiary/10 border-tertiary/20 text-tertiary hover:bg-tertiary/20 active:scale-95'}`}
-                    >
-                        <span className="material-symbols-outlined text-sm">play_arrow</span> START
-                    </button>
-
-                    {/* PAUSE/RESUME - clickable when running */}
-                    <button
-                        onClick={() => sendCommand({ action: 'pause' })}
-                        disabled={!isRunning}
-                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${!isRunning 
-                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
-                            : status === 'PAUSED'
-                                ? 'bg-secondary text-surface border-secondary active:scale-95' 
-                                : 'bg-secondary/10 border-secondary/20 text-secondary hover:bg-secondary/20 active:scale-95'}`}
-                    >
-                        <span className="material-symbols-outlined text-sm">{status === 'PAUSED' ? 'play_arrow' : 'pause'}</span>
-                        {status === 'PAUSED' ? 'RESUME' : 'PAUSE'}
-                    </button>
-
-                    {/* FORCE END - clickable when running */}
-                    <button
-                        onClick={() => sendCommand({ action: 'force_end' })}
-                        disabled={!isRunning}
-                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${!isRunning 
-                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
-                            : 'bg-[#f59e0b]/10 border-[#f59e0b]/20 text-[#f59e0b] hover:bg-[#f59e0b]/20 active:scale-95'}`}
-                    >
-                        <span className="material-symbols-outlined text-sm">stop_circle</span> FORCE END
-                    </button>
-
-                    {/* RESET - always clickable */}
-                    <button
-                        onClick={() => sendCommand({ action: 'reset' })}
-                        className="flex items-center gap-2 px-4 py-1.5 rounded-full bg-error/10 border border-error/20 text-error text-xs font-bold hover:bg-error/20 transition-all active:scale-95"
-                    >
-                        <span className="material-symbols-outlined text-sm">restart_alt</span> RESET
-                    </button>
-                </div>
-
-                <div className="flex items-center gap-2 ml-4">
-                    <div className={`w-2 h-2 rounded-full animate-pulse shadow-[0_0_8px_#66fa8c] ${isConnected ? 'bg-tertiary' : 'bg-error'}`}></div>
-                    <span className={`text-[10px] font-mono font-bold tracking-widest uppercase ${isConnected ? 'text-tertiary' : 'text-error'}`}>
-                        {isConnected ? 'CONNECTED' : 'DISCONNECTED'}
-                    </span>
-                </div>
-            </div>
-        </header>
-    );
-};
-
-export default TopNavBar;
+import React from 'react';
+import { useApp } from '../context/AppContext';
+
+const TopNavBar = () => {
+    const { sessionData, isConnected, sendCommand } = useApp();
+    
+    const status = sessionData?.status || 'STANDBY';
+    const isRunning = sessionData?.active && status !== 'COMPLETED';
+    
+    const isStandby = status === 'STANDBY' || status === 'READY';
+    
+    return (
+        <header className="fixed top-0 w-full z-50 flex justify-between items-center px-6 h-16 bg-surface/60 backdrop-blur-xl border-b border-primary/10 shadow-[0_0_40px_rgba(0,212,255,0.04)]">
+            <div className="flex items-center gap-8">
+                <span className="text-2xl font-black tracking-tighter text-primary font-headline">NEXUS</span>
+                <div className="h-8 w-px bg-outline-variant/20 hidden md:block"></div>
+                <div className="hidden md:flex flex-col">
+                    <span className="text-[10px] font-mono text-outline-variant tracking-widest uppercase">System Status</span>
+                    <div className="text-sm font-mono text-tertiary">{status}</div>
+                </div>
+            </div>
+
+            <div className="flex items-center gap-6">
+                <div className="flex gap-2">
+                    {/* START - clickable when standby */}
+                    <button
+                        onClick={() => sendCommand({ action: 'start' })}
+                        disabled={isRunning}
+                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${isRunning 
+                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
+                            : 'bg-tertiary/10 border-tertiary/20 text-tertiary hover:bg-tertiary/20 active:scale-95'}`}
+                    >
+                        <span className="material-symbols-outlined text-sm">play_arrow</span> START
+                    </button>
+
+                    {/* PAUSE/RESUME - clickable when running */}
+                    <button
+                        onClick={() => sendCommand({ action: 'pause' })}
+                        disabled={!isRunning}
+                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${!isRunning 
+                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
+                            : status === 'PAUSED'
+                                ? 'bg-secondary text-surface border-secondary active:scale-95' 
+                                : 'bg-secondary/10 border-secondary/20 text-secondary hover:bg-secondary/20 active:scale-95'}`}
+                    >
+                        <span className="material-symbols-outlined text-sm">{status === 'PAUSED' ? 'play_arrow' : 'pause'}</span>
+                        {status === 'PAUSED' ? 'RESUME' : 'PAUSE'}
+                    </button>
+
+                    {/* FORCE END - clickable when running */}
+                    <button
+                        onClick={() => sendCommand({ action: 'force_end' })}
+                        disabled={!isRunning}
+                        className={`flex items-center gap-2 px-4 py-1.5 rounded-full border text-xs font-bold transition-all ${!isRunning 
+                            ? 'bg-surface-container text-slate-600 border-slate-700 cursor-not-allowed' 
+                            : 'bg-[#f59e0b]/10 border-[#f59e0b]/20 text-[#f59e0b] hover:bg-[#f59e0b]/20 active:scale-95'}`}
+                    >
+                        <span className="material-symbols-outlined text-sm">stop_circle</span> FORCE END
+                    </button>
+
+                    {/* RESET - always clickable */}
+                    <button
+                        onClick={() => sendCommand({ action: 'reset' })}
+                        className="flex items-center gap-2 px-4 py-1.5 rounded-full bg-error/10 border border-error/20 text-error text-xs font-bold hover:bg-error/20 transition-all active:scale-95"
+                    >
+                        <span className="material-symbols-outlined text-sm">restart_alt</span> RESET
+                    </button>
+                </div>
+
+                <div className="flex items-center gap-2 ml-4">
+                    <div className={`w-2 h-2 rounded-full animate-pulse shadow-[0_0_8px_#66fa8c] ${isConnected ? 'bg-tertiary' : 'bg-error'}`}></div>
+                    <span className={`text-[10px] font-mono font-bold tracking-widest uppercase ${isConnected ? 'text-tertiary' : 'text-error'}`}>
+                        {isConnected ? 'CONNECTED' : 'DISCONNECTED'}
+                    </span>
+                </div>
+            </div>
+        </header>
+    );
+};
+
+export default TopNavBar;

frontend/src/context/AppContext.jsx

@@ -1,48 +1,48 @@
-import React, { createContext, useContext, useState, useEffect, useMemo } from 'react';
-import { config } from '../config';
-import useWebSocket from '../hooks/useWebSocket';
-const AppContext = createContext();
-
-export const AppProvider = ({ children }) => {
-    const [globalMaxSteps, setGlobalMaxSteps] = useState(30);
-    const [simulationSeconds, setSimulationSeconds] = useState(0);
-    const { gameState, isConnected, sendCommand } = useWebSocket(config.WS_URL);
-
-    useEffect(() => {
-        const status = gameState?.status;
-        if (status === 'STANDBY' || status === 'COMPLETED') {
-            setSimulationSeconds(0);
-            return;
-        }
-        if (status === 'PAUSED') {
-            return;
-        }
-        const interval = setInterval(() => {
-            setSimulationSeconds(s => s + 1);
-        }, 1000);
-        return () => clearInterval(interval);
-    }, [gameState?.status]);
-
-    const value = useMemo(() => ({
-        sessionData: gameState,
-        isConnected,
-        sendCommand,
-        globalMaxSteps,
-        setGlobalMaxSteps,
-        simulationSeconds
-    }), [gameState, isConnected, sendCommand, globalMaxSteps, simulationSeconds]);
-
-    return (
-        <AppContext.Provider value={value}>
-            {children}
-        </AppContext.Provider>
-    );
-};
-
-export const useApp = () => {
-    const context = useContext(AppContext);
-    if (!context) {
-        throw new Error('useApp must be used within an AppProvider');
-    }
-    return context;
-};
+import React, { createContext, useContext, useState, useEffect, useMemo } from 'react';
+import { config } from '../config';
+import useWebSocket from '../hooks/useWebSocket';
+const AppContext = createContext();
+
+export const AppProvider = ({ children }) => {
+    const [globalMaxSteps, setGlobalMaxSteps] = useState(30);
+    const [simulationSeconds, setSimulationSeconds] = useState(0);
+    const { gameState, isConnected, sendCommand } = useWebSocket(config.WS_URL);
+
+    useEffect(() => {
+        const status = gameState?.status;
+        if (status === 'STANDBY' || status === 'COMPLETED') {
+            setSimulationSeconds(0);
+            return;
+        }
+        if (status === 'PAUSED') {
+            return;
+        }
+        const interval = setInterval(() => {
+            setSimulationSeconds(s => s + 1);
+        }, 1000);
+        return () => clearInterval(interval);
+    }, [gameState?.status]);
+
+    const value = useMemo(() => ({
+        sessionData: gameState,
+        isConnected,
+        sendCommand,
+        globalMaxSteps,
+        setGlobalMaxSteps,
+        simulationSeconds
+    }), [gameState, isConnected, sendCommand, globalMaxSteps, simulationSeconds]);
+
+    return (
+        <AppContext.Provider value={value}>
+            {children}
+        </AppContext.Provider>
+    );
+};
+
+export const useApp = () => {
+    const context = useContext(AppContext);
+    if (!context) {
+        throw new Error('useApp must be used within an AppProvider');
+    }
+    return context;
+};

frontend/src/hooks/useWebSocket.js

@@ -1,214 +1,214 @@
-import { useState, useEffect, useCallback, useRef } from 'react';
-
-const useWebSocket = (url) => {
-    const [events, setEvents] = useState([]);
-    const [gameState, setGameState] = useState({
-        scenario: null,
-        active: false,
-        status: 'AWAITING_OBJECTIVE',
-        step: 0,
-        reward: 0,
-        cumulativeReward: 0,
-        agent_a_model: '',
-        agent_b_model: '',
-        agents: {
-            agent_a: { status: 'STANDBY', messages: [] },
-            agent_b: { status: 'STANDBY', messages: [] }
-        },
-        clues_found: [],
-        rewardBreakdown: {},
-        rewardHistory: []
-    });
-
-    const [isConnected, setIsConnected] = useState(false);
-    const [error, setError] = useState(null);
-    const socketRef = useRef(null);
-
-    useEffect(() => {
-        socketRef.current = new WebSocket(url);
-
-        socketRef.current.onopen = () => setIsConnected(true);
-
-        socketRef.current.onmessage = (event) => {
-            const data = JSON.parse(event.data);
-            setEvents(prev => [...prev, data]);
-
-            setGameState(prev => {
-                let current = { ...prev };
-
-                if (data.type === 'episode_start') {
-                    return {
-                        ...current,
-                        scenario: data.scenario,
-                        active: true,
-                        status: 'INVESTIGATING',
-                        step: 0,
-                        reward: 0,
-                        cumulativeReward: 0,
-                        clues_found: [],
-                        agent_a_model: data.agent_a_model || current.agent_a_model,
-                        agent_b_model: data.agent_b_model || current.agent_b_model,
-                        agents: {
-                            agent_a: { status: 'ACTIVE', messages: [] },
-                            agent_b: { status: 'ACTIVE', messages: [] }
-                        }
-                    };
-                }
-
-                const newState = { ...current };
-
-                if (data.step !== undefined) {
-                    newState.step = data.step;
-                }
-
-                if (data.type === 'agent_partial') {
-                    const agentId = data.agent_id;
-                    const agents = { ...newState.agents };
-                    const agentReference = agents[agentId];
-                    if (agentReference) {
-                        const agent = { ...agentReference };
-                        const messages = [...(agent.messages || [])];
-                        const lastMsg = messages[messages.length - 1];
-                        if (lastMsg && lastMsg.type === 'message' && lastMsg.partial) {
-                            messages[messages.length - 1] = { ...lastMsg, content: data.full_message };
-                        } else {
-                            messages.push({
-                                type: 'message',
-                                content: data.full_message,
-                                partial: true
-                            });
-                        }
-                        agent.messages = messages;
-                        agents[agentId] = agent;
-                        newState.agents = agents;
-                    }
-                }
-
-                if (data.type === 'agent_message') {
-                    const agentId = data.agent_id;
-                    const agents = { ...newState.agents };
-                    const agentReference = agents[agentId];
-                    if (agentReference) {
-                        const agent = { ...agentReference };
-                        const messages = [...(agent.messages || [])];
-                        const lastMsg = messages[messages.length - 1];
-                        if (lastMsg && lastMsg.partial) {
-                            messages[messages.length - 1] = { ...lastMsg, content: data.message, partial: undefined };
-                        } else {
-                            messages.push({
-                                type: 'message',
-                                content: data.message
-                            });
-                        }
-                        agent.messages = messages;
-                        agents[agentId] = agent;
-                        newState.agents = agents;
-                    }
-                }
-
-                if (data.status === 'READY') {
-                    newState.status = 'READY_TO_INJECT';
-                    newState.active = false;
-                    newState.agents = {
-                        agent_a: { ...newState.agents.agent_a, messages: [] },
-                        agent_b: { ...newState.agents.agent_b, messages: [] }
-                    };
-                }
-
-                if (data.type === 'system_status') {
-                    if (data.paused !== undefined) {
-                        newState.status = data.paused ? 'PAUSED' : 'INVESTIGATING';
-                    }
-                    if (data.status) {
-                        newState.status = data.status;
-                    }
-                    if (data.active !== undefined) {
-                        newState.active = data.active;
-                    }
-                }
-
-                if (data.type === 'tool_call') {
-                    const agentId = data.agent_id;
-                    const agents = { ...newState.agents };
-                    if (agents[agentId]) {
-                        const agent = { ...agents[agentId], messages: [...(agents[agentId].messages || [])] };
-                        agent.messages.push({
-                            type: 'tool_call',
-                            tool_name: data.tool_name,
-                            params: data.params
-                        });
-                        agents[agentId] = agent;
-                        newState.agents = agents;
-                    }
-                }
-
-                if (data.type === 'tool_result') {
-                    const agents = { ...newState.agents };
-                    const agentAReference = agents.agent_a;
-                    if (agentAReference) {
-                        const agentA = { ...agentAReference, messages: [...(agentAReference.messages || [])] };
-                        agentA.messages.push({
-                            type: 'tool_result',
-                            tool_name: data.tool_name,
-                            result: data.result,
-                            success: data.success
-                        });
-                        agents.agent_a = agentA;
-                        newState.agents = agents;
-                    }
-
-                    // Simple heuristic for clues if not sent explicitly
-                    const res = data.result?.toLowerCase() || '';
-                    if (res.includes('error') || res.includes('anomaly') || res.includes('warning') || res.includes('degraded') || data.tool_name === 'propose_fix') {
-                        const currentClues = newState.clues_found || [];
-                        if (!currentClues.includes(data.result)) {
-                            newState.clues_found = [...currentClues, data.result];
-                        }
-                    }
-                }
-
-                if (data.type === 'reward_update') {
-                    newState.reward = data.reward;
-                    newState.cumulativeReward = data.cumulative;
-                    newState.rewardBreakdown = data.breakdown || {};
-                    newState.rewardHistory = [...(newState.rewardHistory || []), data.reward];
-                }
-
-                if (data.type === 'episode_end') {
-                    newState.active = false;
-                    newState.status = 'COMPLETED';
-                    newState.step = data.steps_taken || newState.step;
-                    newState.cumulativeReward = data.final_score !== undefined ? data.final_score : newState.cumulativeReward;
-                    newState.finalScore = data.final_score;
-                    newState.success = data.success;
-                    newState.fixVerified = data.fix_verified;
-                    if (data.clues_found) newState.clues_found = data.clues_found;
-                    if (data.reward_history) newState.rewardHistory = data.reward_history;
-                    if (data.final_breakdown) newState.rewardBreakdown = data.final_breakdown;
-
-                    newState.agents = {
-                        agent_a: { ...newState.agents.agent_a, status: 'STANDBY' },
-                        agent_b: { ...newState.agents.agent_b, status: 'STANDBY' }
-                    };
-                }
-
-                return newState;
-            });
-        };
-
-        socketRef.current.onerror = (err) => setError(err);
-        socketRef.current.onclose = () => setIsConnected(false);
-
-        return () => socketRef.current.close();
-    }, [url]);
-
-    const sendCommand = useCallback((command) => {
-        if (socketRef.current && isConnected) {
-            socketRef.current.send(JSON.stringify(command));
-        }
-    }, [isConnected]);
-
-    return { events, gameState, isConnected, error, sendCommand };
-};
-
-export default useWebSocket;
+import { useState, useEffect, useCallback, useRef } from 'react';
+
+const useWebSocket = (url) => {
+    const [events, setEvents] = useState([]);
+    const [gameState, setGameState] = useState({
+        scenario: null,
+        active: false,
+        status: 'AWAITING_OBJECTIVE',
+        step: 0,
+        reward: 0,
+        cumulativeReward: 0,
+        agent_a_model: '',
+        agent_b_model: '',
+        agents: {
+            agent_a: { status: 'STANDBY', messages: [] },
+            agent_b: { status: 'STANDBY', messages: [] }
+        },
+        clues_found: [],
+        rewardBreakdown: {},
+        rewardHistory: []
+    });
+
+    const [isConnected, setIsConnected] = useState(false);
+    const [error, setError] = useState(null);
+    const socketRef = useRef(null);
+
+    useEffect(() => {
+        socketRef.current = new WebSocket(url);
+
+        socketRef.current.onopen = () => setIsConnected(true);
+
+        socketRef.current.onmessage = (event) => {
+            const data = JSON.parse(event.data);
+            setEvents(prev => [...prev, data]);
+
+            setGameState(prev => {
+                let current = { ...prev };
+
+                if (data.type === 'episode_start') {
+                    return {
+                        ...current,
+                        scenario: data.scenario,
+                        active: true,
+                        status: 'INVESTIGATING',
+                        step: 0,
+                        reward: 0,
+                        cumulativeReward: 0,
+                        clues_found: [],
+                        agent_a_model: data.agent_a_model || current.agent_a_model,
+                        agent_b_model: data.agent_b_model || current.agent_b_model,
+                        agents: {
+                            agent_a: { status: 'ACTIVE', messages: [] },
+                            agent_b: { status: 'ACTIVE', messages: [] }
+                        }
+                    };
+                }
+
+                const newState = { ...current };
+
+                if (data.step !== undefined) {
+                    newState.step = data.step;
+                }
+
+                if (data.type === 'agent_partial') {
+                    const agentId = data.agent_id;
+                    const agents = { ...newState.agents };
+                    const agentReference = agents[agentId];
+                    if (agentReference) {
+                        const agent = { ...agentReference };
+                        const messages = [...(agent.messages || [])];
+                        const lastMsg = messages[messages.length - 1];
+                        if (lastMsg && lastMsg.type === 'message' && lastMsg.partial) {
+                            messages[messages.length - 1] = { ...lastMsg, content: data.full_message };
+                        } else {
+                            messages.push({
+                                type: 'message',
+                                content: data.full_message,
+                                partial: true
+                            });
+                        }
+                        agent.messages = messages;
+                        agents[agentId] = agent;
+                        newState.agents = agents;
+                    }
+                }
+
+                if (data.type === 'agent_message') {
+                    const agentId = data.agent_id;
+                    const agents = { ...newState.agents };
+                    const agentReference = agents[agentId];
+                    if (agentReference) {
+                        const agent = { ...agentReference };
+                        const messages = [...(agent.messages || [])];
+                        const lastMsg = messages[messages.length - 1];
+                        if (lastMsg && lastMsg.partial) {
+                            messages[messages.length - 1] = { ...lastMsg, content: data.message, partial: undefined };
+                        } else {
+                            messages.push({
+                                type: 'message',
+                                content: data.message
+                            });
+                        }
+                        agent.messages = messages;
+                        agents[agentId] = agent;
+                        newState.agents = agents;
+                    }
+                }
+
+                if (data.status === 'READY') {
+                    newState.status = 'READY_TO_INJECT';
+                    newState.active = false;
+                    newState.agents = {
+                        agent_a: { ...newState.agents.agent_a, messages: [] },
+                        agent_b: { ...newState.agents.agent_b, messages: [] }
+                    };
+                }
+
+                if (data.type === 'system_status') {
+                    if (data.paused !== undefined) {
+                        newState.status = data.paused ? 'PAUSED' : 'INVESTIGATING';
+                    }
+                    if (data.status) {
+                        newState.status = data.status;
+                    }
+                    if (data.active !== undefined) {
+                        newState.active = data.active;
+                    }
+                }
+
+                if (data.type === 'tool_call') {
+                    const agentId = data.agent_id;
+                    const agents = { ...newState.agents };
+                    if (agents[agentId]) {
+                        const agent = { ...agents[agentId], messages: [...(agents[agentId].messages || [])] };
+                        agent.messages.push({
+                            type: 'tool_call',
+                            tool_name: data.tool_name,
+                            params: data.params
+                        });
+                        agents[agentId] = agent;
+                        newState.agents = agents;
+                    }
+                }
+
+                if (data.type === 'tool_result') {
+                    const agents = { ...newState.agents };
+                    const agentAReference = agents.agent_a;
+                    if (agentAReference) {
+                        const agentA = { ...agentAReference, messages: [...(agentAReference.messages || [])] };
+                        agentA.messages.push({
+                            type: 'tool_result',
+                            tool_name: data.tool_name,
+                            result: data.result,
+                            success: data.success
+                        });
+                        agents.agent_a = agentA;
+                        newState.agents = agents;
+                    }
+
+                    // Simple heuristic for clues if not sent explicitly
+                    const res = data.result?.toLowerCase() || '';
+                    if (res.includes('error') || res.includes('anomaly') || res.includes('warning') || res.includes('degraded') || data.tool_name === 'propose_fix') {
+                        const currentClues = newState.clues_found || [];
+                        if (!currentClues.includes(data.result)) {
+                            newState.clues_found = [...currentClues, data.result];
+                        }
+                    }
+                }
+
+                if (data.type === 'reward_update') {
+                    newState.reward = data.reward;
+                    newState.cumulativeReward = data.cumulative;
+                    newState.rewardBreakdown = data.breakdown || {};
+                    newState.rewardHistory = [...(newState.rewardHistory || []), data.reward];
+                }
+
+                if (data.type === 'episode_end') {
+                    newState.active = false;
+                    newState.status = 'COMPLETED';
+                    newState.step = data.steps_taken || newState.step;
+                    newState.cumulativeReward = data.final_score !== undefined ? data.final_score : newState.cumulativeReward;
+                    newState.finalScore = data.final_score;
+                    newState.success = data.success;
+                    newState.fixVerified = data.fix_verified;
+                    if (data.clues_found) newState.clues_found = data.clues_found;
+                    if (data.reward_history) newState.rewardHistory = data.reward_history;
+                    if (data.final_breakdown) newState.rewardBreakdown = data.final_breakdown;
+
+                    newState.agents = {
+                        agent_a: { ...newState.agents.agent_a, status: 'STANDBY' },
+                        agent_b: { ...newState.agents.agent_b, status: 'STANDBY' }
+                    };
+                }
+
+                return newState;
+            });
+        };
+
+        socketRef.current.onerror = (err) => setError(err);
+        socketRef.current.onclose = () => setIsConnected(false);
+
+        return () => socketRef.current.close();
+    }, [url]);
+
+    const sendCommand = useCallback((command) => {
+        if (socketRef.current && isConnected) {
+            socketRef.current.send(JSON.stringify(command));
+        }
+    }, [isConnected]);
+
+    return { events, gameState, isConnected, error, sendCommand };
+};
+
+export default useWebSocket;

openenv.yaml

@@ -1,59 +1,59 @@
-name: nexus-incident-investigation
-version: "1.0.0"
-tags: ["openenv"]
-description: >
-  NEXUS — Dual Agent Incident Investigation Environment.
-  Two AI agents collaborate to investigate real-world system incidents.
-  Agent A (Investigator) proposes hypotheses and calls tools.
-  Agent B (Validator) challenges claims and verifies fixes.
-  Together they identify root causes across software, business-process,
-  and cascade-system failure scenarios.
-
-tasks:
-  - name: software-incident
-    description: Single-service software bug causing user-facing errors
-    difficulty: easy
-    max_steps: 8
-    grader: scenarios/graders/easy_grader.py
-
-  - name: business-process-failure
-    description: Multi-team process breakdown with misleading red-herrings
-    difficulty: medium
-    max_steps: 8
-    grader: scenarios/graders/medium_grader.py
-
-  - name: cascade-system-failure
-    description: Multi-system cascade failure with misleading logs
-    difficulty: hard
-    max_steps: 8
-    grader: scenarios/graders/hard_grader.py
-
-action_space:
-  type: text
-  description: Free-form natural language message with optional TOOL: calls
-
-observation_space:
-  type: structured
-  fields:
-    scenario_description: string
-    scenario_context: string
-    partner_message: string
-    tool_results: list
-    clues_found: list
-    investigation_stage: string
-    round: integer
-    available_tools: list
-
-reward_range: [0.0, 1.0]
-reward_description: >
-  Dynamically computed from semantic similarity of hypothesis to root-cause,
-  tool quality, fix correctness, and investigation efficiency.
-
-inference_script: inference.py
-entry_point: backend/main.py
-docker_port: 7860
-
-baseline_scores:
-  software-incident: 0.88
-  business-process-failure: 0.72
-  cascade-system-failure: 0.48
+name: nexus-incident-investigation
+version: "1.0.0"
+tags: ["openenv"]
+description: >
+  NEXUS — Dual Agent Incident Investigation Environment.
+  Two AI agents collaborate to investigate real-world system incidents.
+  Agent A (Investigator) proposes hypotheses and calls tools.
+  Agent B (Validator) challenges claims and verifies fixes.
+  Together they identify root causes across software, business-process,
+  and cascade-system failure scenarios.
+
+tasks:
+  - name: software-incident
+    description: Single-service software bug causing user-facing errors
+    difficulty: easy
+    max_steps: 8
+    grader: scenarios/graders/easy_grader.py
+
+  - name: business-process-failure
+    description: Multi-team process breakdown with misleading red-herrings
+    difficulty: medium
+    max_steps: 8
+    grader: scenarios/graders/medium_grader.py
+
+  - name: cascade-system-failure
+    description: Multi-system cascade failure with misleading logs
+    difficulty: hard
+    max_steps: 8
+    grader: scenarios/graders/hard_grader.py
+
+action_space:
+  type: text
+  description: Free-form natural language message with optional TOOL: calls
+
+observation_space:
+  type: structured
+  fields:
+    scenario_description: string
+    scenario_context: string
+    partner_message: string
+    tool_results: list
+    clues_found: list
+    investigation_stage: string
+    round: integer
+    available_tools: list
+
+reward_range: [0.0, 1.0]
+reward_description: >
+  Dynamically computed from semantic similarity of hypothesis to root-cause,
+  tool quality, fix correctness, and investigation efficiency.
+
+inference_script: inference.py
+entry_point: backend/main.py
+docker_port: 7860
+
+baseline_scores:
+  software-incident: 0.88
+  business-process-failure: 0.72
+  cascade-system-failure: 0.48

pyproject.toml

@@ -15,6 +15,7 @@ dependencies = [
     "httpx>=0.24.0",
     "openai>=1.0.0",
     "psutil>=5.9.0",
+    "openenv-core>=0.2.0",
 ]
 
 [project.scripts]

setup.bat

@@ -1,66 +1,66 @@
-@echo off
-echo ==============================================================
-echo NEXUS Incident Investigation Environment Setup
-echo ==============================================================
-echo.
-
-REM Check Python
-python --version >nul 2>&1
-if %errorlevel% neq 0 (
-    echo [ERROR] Python is not installed or not in PATH!
-    pause
-    exit /b
-)
-
-REM Check npm
-npm --version >nul 2>&1
-if %errorlevel% neq 0 (
-    echo [ERROR] Node.js/npm is not installed or not in PATH!
-    pause
-    exit /b
-)
-
-echo [1/3] Setting up Backend Virtual Environment...
-python -m venv backend\venv
-call backend\venv\Scripts\activate.bat
-pip install -r backend\requirements.txt
-
-echo.
-echo [2/3] Setting up Frontend Dependencies...
-cd frontend
-call npm install
-cd ..
-
-echo.
-echo [3/4] Pulling Required LLM Models (Ollama)...
-echo --------------------------------------------------------------
-echo This will ensure you have the correct models for the simulation.
-echo 1. microsoft/Phi-3-mini-4k-instruct (Investigator)
-echo 2. Qwen/Qwen2.5-1.5B-Instruct       (Validator)
-echo 3. all-minilm                      (Reward Engine)
-echo.
-set /p PULL_MODELS="Do you want to pull these models now? (y/n): "
-if /i "%PULL_MODELS%"=="y" (
-    echo [Pulling Phi-3...]
-    ollama pull phi3:mini
-    echo [Pulling Qwen-1.5B...]
-    ollama pull qwen2.5:1.5b
-    echo [Pulling all-minilm...]
-    ollama pull all-minilm
-) else (
-    echo Skipping model pull. Ensure you pull them manually later.
-)
-
-echo.
-echo [4/4] Validating OpenEnv Compliance...
-call backend\venv\Scripts\python.exe openenv_validator.py
-
-echo.
-echo ==============================================================
-echo SETUP COMPLETE!
-echo.
-echo To run locally:
-echo 1. Start UI:    cd frontend ^& npm run dev
-echo 2. Start API:   cd backend ^& venv\Scripts\python main.py
-echo ==============================================================
-pause
+@echo off
+echo ==============================================================
+echo NEXUS Incident Investigation Environment Setup
+echo ==============================================================
+echo.
+
+REM Check Python
+python --version >nul 2>&1
+if %errorlevel% neq 0 (
+    echo [ERROR] Python is not installed or not in PATH!
+    pause
+    exit /b
+)
+
+REM Check npm
+npm --version >nul 2>&1
+if %errorlevel% neq 0 (
+    echo [ERROR] Node.js/npm is not installed or not in PATH!
+    pause
+    exit /b
+)
+
+echo [1/3] Setting up Backend Virtual Environment...
+python -m venv backend\venv
+call backend\venv\Scripts\activate.bat
+pip install -r backend\requirements.txt
+
+echo.
+echo [2/3] Setting up Frontend Dependencies...
+cd frontend
+call npm install
+cd ..
+
+echo.
+echo [3/4] Pulling Required LLM Models (Ollama)...
+echo --------------------------------------------------------------
+echo This will ensure you have the correct models for the simulation.
+echo 1. microsoft/Phi-3-mini-4k-instruct (Investigator)
+echo 2. Qwen/Qwen2.5-1.5B-Instruct       (Validator)
+echo 3. all-minilm                      (Reward Engine)
+echo.
+set /p PULL_MODELS="Do you want to pull these models now? (y/n): "
+if /i "%PULL_MODELS%"=="y" (
+    echo [Pulling Phi-3...]
+    ollama pull phi3:mini
+    echo [Pulling Qwen-1.5B...]
+    ollama pull qwen2.5:1.5b
+    echo [Pulling all-minilm...]
+    ollama pull all-minilm
+) else (
+    echo Skipping model pull. Ensure you pull them manually later.
+)
+
+echo.
+echo [4/4] Validating OpenEnv Compliance...
+call backend\venv\Scripts\python.exe openenv_validator.py
+
+echo.
+echo ==============================================================
+echo SETUP COMPLETE!
+echo.
+echo To run locally:
+echo 1. Start UI:    cd frontend ^& npm run dev
+echo 2. Start API:   cd backend ^& venv\Scripts\python main.py
+echo ==============================================================
+pause

setup.sh

@@ -1,42 +1,42 @@
-#!/bin/bash
-
-echo "=============================================================="
-echo "NEXUS Incident Investigation Environment Setup"
-echo "=============================================================="
-echo ""
-
-# Check Python
-if ! command -v python3 &> /dev/null; then
-    echo "[ERROR] python3 is not installed or not in PATH!"
-    exit 1
-fi
-
-# Check npm
-if ! command -v npm &> /dev/null; then
-    echo "[ERROR] npm is not installed or not in PATH!"
-    exit 1
-fi
-
-echo "[1/3] Setting up Backend Virtual Environment..."
-python3 -m venv backend/venv
-source backend/venv/bin/activate
-pip install -r backend/requirements.txt
-
-echo ""
-echo "[2/3] Setting up Frontend Dependencies..."
-cd frontend
-npm install
-cd ..
-
-echo ""
-echo "[3/3] Validating OpenEnv Compliance..."
-backend/venv/bin/python openenv_validator.py
-
-echo ""
-echo "=============================================================="
-echo "SETUP COMPLETE!"
-echo ""
-echo "To run locally without Docker:"
-echo "1. Start UI:    cd frontend && npm run dev"
-echo "2. Start API:   cd backend && venv/bin/uvicorn main:app --reload"
-echo "=============================================================="
+#!/bin/bash
+
+echo "=============================================================="
+echo "NEXUS Incident Investigation Environment Setup"
+echo "=============================================================="
+echo ""
+
+# Check Python
+if ! command -v python3 &> /dev/null; then
+    echo "[ERROR] python3 is not installed or not in PATH!"
+    exit 1
+fi
+
+# Check npm
+if ! command -v npm &> /dev/null; then
+    echo "[ERROR] npm is not installed or not in PATH!"
+    exit 1
+fi
+
+echo "[1/3] Setting up Backend Virtual Environment..."
+python3 -m venv backend/venv
+source backend/venv/bin/activate
+pip install -r backend/requirements.txt
+
+echo ""
+echo "[2/3] Setting up Frontend Dependencies..."
+cd frontend
+npm install
+cd ..
+
+echo ""
+echo "[3/3] Validating OpenEnv Compliance..."
+backend/venv/bin/python openenv_validator.py
+
+echo ""
+echo "=============================================================="
+echo "SETUP COMPLETE!"
+echo ""
+echo "To run locally without Docker:"
+echo "1. Start UI:    cd frontend && npm run dev"
+echo "2. Start API:   cd backend && venv/bin/uvicorn main:app --reload"
+echo "=============================================================="

tests/test_environment.py

@@ -1,35 +1,35 @@
-import pytest
-import asyncio
-from core.environment import NexusEnvironment
-
-@pytest.mark.asyncio
-async def test_env_reset():
-    env = NexusEnvironment()
-    obs = await env.reset(task="software-incident")
-    assert obs.scenario_description != ""
-    assert "503" in str(obs.scenario_description).lower() or "rate limit" in str(obs.scenario_description).lower()
-    assert env.active_episode is not None
-
-@pytest.mark.asyncio
-async def test_env_step():
-    env = NexusEnvironment()
-    await env.reset(task="software-incident")
-    
-    from api.schemas.action import NexusAction
-    action = NexusAction(
-        agent_id="agent_a",
-        message="Checking Nginx logs",
-        tool_calls=[],
-        confidence=0.5
-    )
-    
-    obs, reward, done, info = await env.step(action)
-    assert reward >= 0.0
-    assert not done
-    assert env.active_episode.steps_taken == 1
-
-@pytest.mark.asyncio
-async def test_invalid_task():
-    env = NexusEnvironment()
-    with pytest.raises(ValueError):
-        await env.reset(task="non-existent-task")
+import pytest
+import asyncio
+from core.environment import NexusEnvironment
+
+@pytest.mark.asyncio
+async def test_env_reset():
+    env = NexusEnvironment()
+    obs = await env.reset(task="software-incident")
+    assert obs.scenario_description != ""
+    assert "503" in str(obs.scenario_description).lower() or "rate limit" in str(obs.scenario_description).lower()
+    assert env.active_episode is not None
+
+@pytest.mark.asyncio
+async def test_env_step():
+    env = NexusEnvironment()
+    await env.reset(task="software-incident")
+    
+    from api.schemas.action import NexusAction
+    action = NexusAction(
+        agent_id="agent_a",
+        message="Checking Nginx logs",
+        tool_calls=[],
+        confidence=0.5
+    )
+    
+    obs, reward, done, info = await env.step(action)
+    assert reward >= 0.0
+    assert not done
+    assert env.active_episode.steps_taken == 1
+
+@pytest.mark.asyncio
+async def test_invalid_task():
+    env = NexusEnvironment()
+    with pytest.raises(ValueError):
+        await env.reset(task="non-existent-task")

tests/test_reward.py

@@ -1,34 +1,34 @@
-import pytest
-from unittest.mock import patch
-from core.reward_engine import compute_reward
-
-def test_reward_engine_basic():
-    # Mock episode state
-    class MockEpisode:
-        def __init__(self):
-            self.all_messages = ["Hello partner, let's investigate the Nginx 503 error."]
-            self.clues_found = []
-            self.previous_tool_calls = []
-            self.steps_taken = 1
-            self.difficulty = "easy"
-            self.last_partner_message = "What do you see?"
-            self.reward_history = []
-            self.cumulative_reward = 0.0
-
-    ep = MockEpisode()
-    
-    # Mock embeddings to avoid needing a server
-    with patch('core.reward_engine.get_embedding', return_value=[0.1]*384), \
-         patch('core.reward_engine.cos_sim', return_value=0.8):
-         
-        final_score, info = compute_reward(
-            message="I will check the configuration file /etc/nginx/nginx.conf",
-            tool_calls=[],
-            tool_results=[],
-            episode_state=ep,
-            scenario={"root_cause": {"description": "Nginx rate limit"}}
-        )
-    
-        assert 0.0 <= final_score <= 1.0
-        assert "specificity" in info
-        assert "progress" in info
+import pytest
+from unittest.mock import patch
+from core.reward_engine import compute_reward
+
+def test_reward_engine_basic():
+    # Mock episode state
+    class MockEpisode:
+        def __init__(self):
+            self.all_messages = ["Hello partner, let's investigate the Nginx 503 error."]
+            self.clues_found = []
+            self.previous_tool_calls = []
+            self.steps_taken = 1
+            self.difficulty = "easy"
+            self.last_partner_message = "What do you see?"
+            self.reward_history = []
+            self.cumulative_reward = 0.0
+
+    ep = MockEpisode()
+    
+    # Mock embeddings to avoid needing a server
+    with patch('core.reward_engine.get_embedding', return_value=[0.1]*384), \
+         patch('core.reward_engine.cos_sim', return_value=0.8):
+         
+        final_score, info = compute_reward(
+            message="I will check the configuration file /etc/nginx/nginx.conf",
+            tool_calls=[],
+            tool_results=[],
+            episode_state=ep,
+            scenario={"root_cause": {"description": "Nginx rate limit"}}
+        )
+    
+        assert 0.0 <= final_score <= 1.0
+        assert "specificity" in info
+        assert "progress" in info

uv.lock

@@ -6,41 +6,6 @@ name = "nexus-ai"
 version = "1.0.0"
 source = { editable = "." }
 
-[[package]]
-name = "fastapi"
-version = "0.115.0"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "uvicorn"
-version = "0.32.0"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "pydantic"
-version = "2.10.0"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "python-dotenv"
-version = "1.0.1"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "httpx"
-version = "0.28.0"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "openai"
-version = "1.58.0"
-source = { registry = "https://pypi.org/simple" }
-
-[[package]]
-name = "psutil"
-version = "6.1.0"
-source = { registry = "https://pypi.org/simple" }
-
 [[package]]
 name = "openenv-core"
 version = "0.2.0"