Datasets:
name stringlengths 34 34 | instruction stringlengths 955 2.5k | description stringlengths 88 104 | keywords listlengths 0 0 | schema_version stringclasses 1
value | config unknown | metadata unknown | files listlengths 5 5 | location stringlengths 30 30 |
|---|---|---|---|---|---|---|---|---|
terminal-desk/task_000004_aeddda76 | You are assisting a storage administrator in managing and securing user uploads. Users frequently upload `.tar.gz` archives to the `/home/user/incoming` directory. However, some of these archives are malicious and contain "zip slip" payloads—files with absolute paths (e.g., `/etc/passwd`) or directory traversal sequenc... | TMax-15K task task_000004_aeddda76 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000004_aeddda76",
"description": "TMax-15K task task_000004_aeddda76 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"veri... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000004_aeddda76/tb2 |
terminal-desk/task_000038_9ef2ca8c | You are a data analyst reviewing a legacy ETL pipeline. We have raw sensor data and the aggregated daily summaries produced by the legacy system. We suspect the legacy system introduced a floating-point calculation error that biases the results.
Your task is to write a Python script that acts as a new ETL verificatio... | TMax-15K task task_000038_9ef2ca8c (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000038_9ef2ca8c",
"description": "TMax-15K task task_000038_9ef2ca8c (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
"verif... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000038_9ef2ca8c/tb2 |
terminal-desk/task_000149_70d2b440 | I have a directory of old CNC project files that I need to organize and analyze. The files are located in `/home/user/projects/cnc_data/` and have a `.rle` extension.
These files were compressed using a custom, primitive Run-Length Encoding (RLE) tool, and the original text inside was encoded in UTF-16LE.
I need you... | TMax-15K task task_000149_70d2b440 (file_operations, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000149_70d2b440",
"description": "TMax-15K task task_000149_70d2b440 (file_operations, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-syst... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000149_70d2b440/tb2 |
terminal-desk/task_000193_cd7fea9b | You are helping clean and aggregate a raw dataset of IoT sensor readings.
The input data is located at `/home/user/raw_sensors.csv` and contains comma-separated values with the following header:
`timestamp,sensor_id,temperature,humidity`
The timestamps are in ISO8601 format (e.g., `2023-10-01T12:34:56Z`).
However, t... | TMax-15K task task_000193_cd7fea9b (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000193_cd7fea9b",
"description": "TMax-15K task task_000193_cd7fea9b (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000193_cd7fea9b/tb2 |
terminal-desk/task_000295_3f39002e | You are assisting a technical writer in organizing and migrating a dump of legacy documentation.
You have been provided with a nested archive located at `/home/user/legacy_docs.zip`.
Please perform the following steps to update the documentation:
1. Extract `/home/user/legacy_docs.zip` into a new directory at `/home/... | TMax-15K task task_000295_3f39002e (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000295_3f39002e",
"description": "TMax-15K task task_000295_3f39002e (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000295_3f39002e/tb2 |
terminal-desk/task_000298_bd10fc23 | You are an AI assistant helping a technical writer organize a messy export of documentation assets.
The writer has a directory of raw files at `/home/user/docs_raw`. Many of the files have incorrect or missing extensions. You need to identify all the actual PNG image files, move them to a designated folder, and create... | TMax-15K task task_000298_bd10fc23 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000298_bd10fc23",
"description": "TMax-15K task task_000298_bd10fc23 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000298_bd10fc23/tb2 |
terminal-desk/task_000359_1bf23249 | You are a compliance analyst tasked with generating an audit trail for a custom authentication service. The legacy Python application is located at `/home/user/app/auth_backend.py`.
The developers claim the authentication flow is secure and that the `get_admin_data` function is properly protected against privilege es... | TMax-15K task task_000359_1bf23249 (security, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000359_1bf23249",
"description": "TMax-15K task task_000359_1bf23249 (security, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifier": ... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000359_1bf23249/tb2 |
terminal-desk/task_000425_efb24a23 | You are a penetration tester analyzing a compromised Linux server. You have discovered a suspicious executable left by an attacker and an encrypted payload. Your objective is to reverse engineer the binary to find a cryptographic key, use it to decrypt the payload, and analyze the resulting TLS certificate.
You must a... | TMax-15K task task_000425_efb24a23 (security, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000425_efb24a23",
"description": "TMax-15K task task_000425_efb24a23 (security, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifier... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000425_efb24a23/tb2 |
terminal-desk/task_000568_82bf0966 | You are acting as a localization engineer. Our automated translation ETL job experienced network issues last week and retried multiple times, resulting in a large file containing duplicate, out-of-order, and unnormalized translation records.
Your task is to build a Python data pipeline to clean, deduplicate, aggregat... | TMax-15K task task_000568_82bf0966 (data_processing, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000568_82bf0966",
"description": "TMax-15K task task_000568_82bf0966 (data_processing, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000568_82bf0966/tb2 |
terminal-desk/task_000661_93996c7e | You are managing a legacy cluster. Periodically, the system dumps configuration state into compressed backup files. You need to extract database credentials from these backups to audit them.
In the directory `/home/user/legacy_configs/`, there are three gzipped configuration files:
- `node_A.cfg.gz`
- `node_B.cfg.gz`
... | TMax-15K task task_000661_93996c7e (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000661_93996c7e",
"description": "TMax-15K task task_000661_93996c7e (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"veri... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000661_93996c7e/tb2 |
terminal-desk/task_000765_32f43201 | You are a data analyst troubleshooting a pipeline. A nightly job dumped a corrupted CSV file containing daily sales totals into `/home/user/sales.csv`.
The file has two columns: `Date` (YYYY-MM-DD) and `Sales`. However, the rows are out of order, and several days are completely missing their `Sales` values.
Using on... | TMax-15K task task_000765_32f43201 (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000765_32f43201",
"description": "TMax-15K task task_000765_32f43201 (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000765_32f43201/tb2 |
terminal-desk/task_000769_80b7eb4c | You are an AI assistant helping a technical writer organize and extract configurations from a set of fragmented, periodically rotated documentation archives.
The writer has several documentation archives located in `/home/user/docs_raw/`. These archives (`part1.tar.gz` and `part2.tar.gz`) contain Markdown files with ... | TMax-15K task task_000769_80b7eb4c (file_operations, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000769_80b7eb4c",
"description": "TMax-15K task task_000769_80b7eb4c (file_operations, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000769_80b7eb4c/tb2 |
terminal-desk/task_000825_83527965 | You are a data analyst tasked with building a reproducible ETL script to clean a noisy dataset of server metrics.
The raw data is located at `/home/user/raw_metrics.csv` and contains four columns: `timestamp`, `cpu_usage`, `memory_usage`, and `response_time`.
Please perform the following operations to clean the data:... | TMax-15K task task_000825_83527965 (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000825_83527965",
"description": "TMax-15K task task_000825_83527965 (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000825_83527965/tb2 |
terminal-desk/task_000894_b88dd812 | You are a red-team operator simulating an attack on a client's web server. You have identified an open redirect vulnerability in the server's login flow. The server takes an obfuscated `redirect_uri` parameter, deobfuscates it, and redirects the user.
You managed to recover the server-side deobfuscation logic. It is ... | TMax-15K task task_000894_b88dd812 (security, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000894_b88dd812",
"description": "TMax-15K task task_000894_b88dd812 (security, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifier": ... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000894_b88dd812/tb2 |
terminal-desk/task_000899_ba9ad6f9 | I am a researcher organizing a departmental collaboration graph dataset. I have two relational CSV files representing a co-authorship network, and I need to map this relational data into an analytical document format using Bash.
The files are located at:
1. `/home/user/nodes.csv` - Contains `node_id,author_name,depart... | TMax-15K task task_000899_ba9ad6f9 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_000899_ba9ad6f9",
"description": "TMax-15K task task_000899_ba9ad6f9 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_000899_ba9ad6f9/tb2 |
terminal-desk/task_001052_30fddd68 | You are helping a technical writer update a legacy documentation archive.
You have been provided with a compressed tar archive located at `/home/user/archive/legacy_docs.tar.gz`. This archive contains thousands of markdown files. The company has recently rebranded from "AcmeCorp" to "NovaCorp". You need to update all... | TMax-15K task task_001052_30fddd68 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001052_30fddd68",
"description": "TMax-15K task task_001052_30fddd68 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001052_30fddd68/tb2 |
terminal-desk/task_001104_34eb0a40 | You are a data analyst working with a stream of sensor data.
You have been provided with a CSV file at `/home/user/raw_sensors.csv`. This file contains data in a "wide" format, where each row represents a `timestamp` and subsequent columns (`S1`, `S2`, `S3`, `S4`, `S5`) represent readings from different sensors.
Un... | TMax-15K task task_001104_34eb0a40 (data_processing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001104_34eb0a40",
"description": "TMax-15K task task_001104_34eb0a40 (data_processing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-syst... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001104_34eb0a40/tb2 |
terminal-desk/task_001177_39a6146d | You are a data analyst working for a manufacturing company. You only have access to standard Linux command-line tools (Bash, awk, grep, etc.) to process some exported CSV files.
You have been given two CSV files representing a Bill of Materials (BOM) and a component catalog:
1. `/home/user/components.csv`
Schema: `... | TMax-15K task task_001177_39a6146d (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001177_39a6146d",
"description": "TMax-15K task task_001177_39a6146d (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001177_39a6146d/tb2 |
terminal-desk/task_001214_a5e2e205 | I am a technical writer organizing some legacy documentation. My predecessor stored several drafts in a custom concatenated text format instead of a standard archive. I need you to extract the files, clean up their names, and package them properly.
Here is what you need to do:
1. You will find a file at `/home/user/le... | TMax-15K task task_001214_a5e2e205 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001214_a5e2e205",
"description": "TMax-15K task task_001214_a5e2e205 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001214_a5e2e205/tb2 |
terminal-desk/task_001218_b5ef52f9 | You are an AI assistant helping a backup administrator recover and archive data after a storage corruption event.
During the recovery process, several files were salvaged into the `/home/user/recovery/inbox/` directory, but their original file extensions were lost or scrambled (they might end in `.dat`, `.tmp`, `.bin... | TMax-15K task task_001218_b5ef52f9 (file_operations, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001218_b5ef52f9",
"description": "TMax-15K task task_001218_b5ef52f9 (file_operations, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
}... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001218_b5ef52f9/tb2 |
terminal-desk/task_001249_f74639d5 | You are acting as a technical assistant to a compliance officer. We are auditing system access across our engineering organization. Access permissions are modeled as a graph in an SQLite database located at `/home/user/compliance.db`. The graph connects Users to Roles, and Roles to Systems.
Our automated bash script `... | TMax-15K task task_001249_f74639d5 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001249_f74639d5",
"description": "TMax-15K task task_001249_f74639d5 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001249_f74639d5/tb2 |
terminal-desk/task_001350_a6e2bdb3 | You are a researcher who needs to organize and analyze experimental dataset files using only shell scripting. You have two datasets collected from a mechanical experiment involving various springs:
- `/home/user/data/measurements.csv` (columns: `id`, `spring_id`, `displacement`)
- `/home/user/data/forces.csv` (columns:... | TMax-15K task task_001350_a6e2bdb3 (data_science, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001350_a6e2bdb3",
"description": "TMax-15K task task_001350_a6e2bdb3 (data_science, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001350_a6e2bdb3/tb2 |
terminal-desk/task_001408_987fa778 | You are helping a technical writer organize a messy archive of compressed documentation drafts.
In the directory `/home/user/drafts/`, there are several gzipped text files named `draft_001.txt.gz`, `draft_002.txt.gz`, and so on. These files are quite large, so you should avoid fully extracting them to disk or loading... | TMax-15K task task_001408_987fa778 (file_operations, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001408_987fa778",
"description": "TMax-15K task task_001408_987fa778 (file_operations, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001408_987fa778/tb2 |
terminal-desk/task_001468_02f89833 | You are a data analyst investigating a series of financial transactions. Some transactions may be causing system deadlocks, and others might have invalid account references due to a data export glitch. You have two files: `/home/user/accounts.csv` and `/home/user/transactions.csv`.
Your tasks:
1. **Schema Validation**... | TMax-15K task task_001468_02f89833 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001468_02f89833",
"description": "TMax-15K task task_001468_02f89833 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001468_02f89833/tb2 |
terminal-desk/task_001617_ba29175b | You are an AI assistant helping a bioinformatics analyst debug a sequence processing pipeline written entirely in Bash and Awk.
We are trying to build a basic De Bruijn graph representation (k-mer transition matrix) to help design primers. The script `/home/user/analyze_kmers.sh` is supposed to read a DNA sequence fro... | TMax-15K task task_001617_ba29175b (scientific_computing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001617_ba29175b",
"description": "TMax-15K task task_001617_ba29175b (scientific_computing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001617_ba29175b/tb2 |
terminal-desk/task_001699_07cef3a1 | You are an MLOps engineer tasked with tracking experiment artifacts and their inference performance.
You have been given a directory of model artifacts at `/home/user/models/` and a benchmarking script at `/home/user/benchmark.py`.
Your goal is to build a short, reproducible Python pipeline script at `/home/user/pi... | TMax-15K task task_001699_07cef3a1 (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001699_07cef3a1",
"description": "TMax-15K task task_001699_07cef3a1 (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001699_07cef3a1/tb2 |
terminal-desk/task_001770_e7e092f5 | As a compliance officer auditing our internal systems, I need to investigate a potential violation of separation of duties. We have a flat log file of access delegations, but no clear map of the overall access network.
I need you to reverse-engineer the access graph from our delegation logs, aggregate some overall ne... | TMax-15K task task_001770_e7e092f5 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001770_e7e092f5",
"description": "TMax-15K task task_001770_e7e092f5 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifi... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001770_e7e092f5/tb2 |
terminal-desk/task_001792_d647aeca | You are acting as a compliance officer auditing a corporate ownership database. We need to identify potential "circular control" structures—specifically, instances where three corporate entities form a closed loop of ownership, which is a pattern often used to obscure true beneficial ownership (similar to how circular ... | TMax-15K task task_001792_d647aeca (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001792_d647aeca",
"description": "TMax-15K task task_001792_d647aeca (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001792_d647aeca/tb2 |
terminal-desk/task_001945_0a0d641b | You are assisting a technical writer in organizing their scattered Markdown documentation into a compressed archive format, while ensuring concurrent safety for future automation.
You are provided with the following environment:
1. A configuration file located at `/home/user/archiver.conf` which contains key-value pai... | TMax-15K task task_001945_0a0d641b (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_001945_0a0d641b",
"description": "TMax-15K task task_001945_0a0d641b (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_001945_0a0d641b/tb2 |
terminal-desk/task_002016_bb429720 | You are a monitoring specialist tasked with setting up a local alert generation script based on firewall and port-forwarding logs.
We have a log file located at `/home/user/fw_logs.txt`.
The log file contains entries in the following format:
`[YYYY-MM-DD HH:MM:SS] <LEVEL>: <MESSAGE> IP:<IP_ADDRESS> PORT:<PORT_NUMBER... | TMax-15K task task_002016_bb429720 (system_administration, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002016_bb429720",
"description": "TMax-15K task task_002016_bb429720 (system_administration, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002016_bb429720/tb2 |
terminal-desk/task_002044_9cb03e7d | You are a localization engineer managing a platform that logs user access to translated strings over time. The platform generates time-series logs in JSON-Lines format. Recently, a bug in the logging system caused Unicode characters in the `text` field to be double-escaped or inconsistently escaped (e.g., `\u00e9` inst... | TMax-15K task task_002044_9cb03e7d (data_processing, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002044_9cb03e7d",
"description": "TMax-15K task task_002044_9cb03e7d (data_processing, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002044_9cb03e7d/tb2 |
terminal-desk/task_002142_7846cdd7 | You are a reliable backend engineer investigating a sudden Out-Of-Memory (OOM) crash in a long-running Python transaction processing service. The service crashed recently due to a massive memory leak triggered by a concurrency bug.
You have been provided with the following directories and files:
1. `/home/user/logs/ap... | TMax-15K task task_002142_7846cdd7 (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002142_7846cdd7",
"description": "TMax-15K task task_002142_7846cdd7 (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
"ver... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002142_7846cdd7/tb2 |
terminal-desk/task_002160_20e23d1b | You are a bioinformatics analyst studying an idealized model of DNA unwinding energy along a synthetic chromosome.
We have generated a sequence energy profile stored in `/home/user/energy_signal.csv`. The file contains two columns: `x` (normalized genomic position from 0 to 10, with a step size of 0.01) and `E` (the m... | TMax-15K task task_002160_20e23d1b (scientific_computing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002160_20e23d1b",
"description": "TMax-15K task task_002160_20e23d1b (scientific_computing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002160_20e23d1b/tb2 |
terminal-desk/task_002181_272aa573 | You are a developer tasked with debugging a failing build in a Rust project.
In your workspace at `/home/user/kepler_solver`, there is a Rust library that solves Kepler's equation ($E - e \sin(E) = M$) for orbital mechanics using the Newton-Raphson method.
Recently, the CI pipeline has started experiencing intermit... | TMax-15K task task_002181_272aa573 (debugging, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002181_272aa573",
"description": "TMax-15K task task_002181_272aa573 (debugging, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasonin... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002181_272aa573/tb2 |
terminal-desk/task_002208_ccc0fae5 | You are a data engineer tasked with building an ETL pipeline to analyze transaction graphs using Bash and standard CLI utilities (`awk`, `jq`, `sort`, etc.).
You have been provided with a raw CSV file at `/home/user/transactions.csv` containing transaction logs. The file has a header and the following columns: `sende... | TMax-15K task task_002208_ccc0fae5 (data_querying, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002208_ccc0fae5",
"description": "TMax-15K task task_002208_ccc0fae5 (data_querying, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002208_ccc0fae5/tb2 |
terminal-desk/task_002361_260fbe95 | You are a data analyst tasked with processing a Bill of Materials (BOM) represented as a directed acyclic graph.
You have been provided with two CSV files in your home directory:
1. `/home/user/nodes.csv`: Contains the nodes of the graph. Columns: `node_id`, `base_cost`, `node_type`.
2. `/home/user/edges.csv`: Contain... | TMax-15K task task_002361_260fbe95 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002361_260fbe95",
"description": "TMax-15K task task_002361_260fbe95 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoni... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002361_260fbe95/tb2 |
terminal-desk/task_002371_aa4c1263 | You are a compliance officer auditing a financial system for anomalous transaction patterns. The engineering team recently discovered that the system can suffer from a "deadlock" state when two users initiate simultaneous transactions to each other at the exact same millisecond.
You have been provided an export of th... | TMax-15K task task_002371_aa4c1263 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002371_aa4c1263",
"description": "TMax-15K task task_002371_aa4c1263 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
"ve... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002371_aa4c1263/tb2 |
terminal-desk/task_002466_0552bfa5 | You are a data engineer tasked with building a lightweight, highly reproducible ETL pipeline using only standard Linux shell tools (Bash, awk, sort, join, etc.).
You have been given two comma-separated data files without headers:
1. `/home/user/logins.csv` contains `user_id`, `login_count`, and `days_inactive`.
2. `/h... | TMax-15K task task_002466_0552bfa5 (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002466_0552bfa5",
"description": "TMax-15K task task_002466_0552bfa5 (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"v... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002466_0552bfa5/tb2 |
terminal-desk/task_002476_2cba6900 | You are helping a computational biology researcher analyze a network dataset.
A SQLite database is located at `/home/user/graph_data.db`. It contains the following schema:
- `nodes`: `(id INT, label TEXT, property TEXT)`
- `relations`: `(src INT, dst INT, type TEXT)`
The researcher needs to extract all instances of ... | TMax-15K task task_002476_2cba6900 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002476_2cba6900",
"description": "TMax-15K task task_002476_2cba6900 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoni... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002476_2cba6900/tb2 |
terminal-desk/task_002490_b7f8e82e | You are an ML engineer preparing training data for a new model.
We have a simple data processing script located at `/home/user/process.py`. It reads a raw CSV dataset from `/home/user/data/raw_data.csv` and outputs a Parquet file to `/home/user/processed.parquet`.
Currently, the downstream training pipeline is cras... | TMax-15K task task_002490_b7f8e82e (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002490_b7f8e82e",
"description": "TMax-15K task task_002490_b7f8e82e (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002490_b7f8e82e/tb2 |
terminal-desk/task_002713_f0648f32 | You are a Data Scientist preparing a text dataset for a Natural Language Processing (NLP) model. You need to build a data preparation pipeline using only Bash shell built-ins and standard POSIX command-line utilities (like `awk`, `sed`, `grep`, `join`, `sort`, `tr`, etc.). Do not use Python, Perl, or any other programm... | TMax-15K task task_002713_f0648f32 (data_science, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002713_f0648f32",
"description": "TMax-15K task task_002713_f0648f32 (data_science, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002713_f0648f32/tb2 |
terminal-desk/task_002727_d910c01d | You are helping a data analyst who is struggling with a buggy bash script.
The analyst wrote a script located at `/home/user/generate_report.sh`. This script imports two CSV files (`/home/user/employees.csv` and `/home/user/departments.csv`) into an SQLite database and attempts to query them to create an organization... | TMax-15K task task_002727_d910c01d (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002727_d910c01d",
"description": "TMax-15K task task_002727_d910c01d (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002727_d910c01d/tb2 |
terminal-desk/task_002744_e3a2ea46 | You are an IT compliance officer auditing a financial transaction database for potential high-risk money movement chains.
You have an SQLite database located at `/home/user/financials.db`. It contains a single table:
`transfers(id INTEGER PRIMARY KEY, source TEXT, target TEXT, amount REAL)`
Your objective is to write... | TMax-15K task task_002744_e3a2ea46 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002744_e3a2ea46",
"description": "TMax-15K task task_002744_e3a2ea46 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002744_e3a2ea46/tb2 |
terminal-desk/task_002959_8f669053 | You are a FinOps analyst tasked with automating the extraction and organization of cloud cost anomalies from daily billing logs.
You have a directory `/home/user/billing_logs/` containing daily CSV files named by date (e.g., `2023-10-01.csv`). Each file has the following columns: `date,service,resource_id,cost`.
Ple... | TMax-15K task task_002959_8f669053 (system_administration, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002959_8f669053",
"description": "TMax-15K task task_002959_8f669053 (system_administration, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002959_8f669053/tb2 |
terminal-desk/task_002962_a50faf04 | You are acting as a Database Administrator. We have inherited an undocumented SQLite database located at `/home/user/company.db`.
Your task is to:
1. Reverse engineer the schema of `/home/user/company.db` to understand the relationships between customers, orders, order items, and products.
2. Write a Python script at... | TMax-15K task task_002962_a50faf04 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002962_a50faf04",
"description": "TMax-15K task task_002962_a50faf04 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002962_a50faf04/tb2 |
terminal-desk/task_002978_eb2c9961 | You are acting as a compliance officer auditing a financial system for suspicious activities. You have been provided with an SQLite database at `/home/user/financial_audit.db` containing a single table `wire_transfers` with the following schema:
`tx_id INTEGER PRIMARY KEY, sender TEXT, receiver TEXT, amount REAL, times... | TMax-15K task task_002978_eb2c9961 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_002978_eb2c9961",
"description": "TMax-15K task task_002978_eb2c9961 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
"ve... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_002978_eb2c9961/tb2 |
terminal-desk/task_003052_49f70213 | You are an AI assistant helping a bioinformatics researcher organize and analyze a dataset of biological interactions.
The researcher has a dataset of biological relationships in a CSV file located at `/home/user/interactions.csv`. The CSV has no header and contains the following columns:
`source_entity`, `source_typ... | TMax-15K task task_003052_49f70213 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003052_49f70213",
"description": "TMax-15K task task_003052_49f70213 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
"v... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003052_49f70213/tb2 |
terminal-desk/task_003086_4d8d54be | You are a data analyst tasked with cleaning an international weather sensor dataset. The raw data is located at `/home/user/sensors_raw.csv`.
The CSV has three columns: `date` (YYYY-MM-DD), `station` (string), and `temp` (float, but some are missing).
The dataset has several issues that you need to fix using a Pytho... | TMax-15K task task_003086_4d8d54be (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003086_4d8d54be",
"description": "TMax-15K task task_003086_4d8d54be (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
"ve... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003086_4d8d54be/tb2 |
terminal-desk/task_003123_578631d1 | You are an AI assistant acting as a localization engineer. You need to consolidate translation files from different platforms into a single unified format.
Currently, our translations are scattered across two files in `/home/user/locales/`:
1. `app_strings.json`: Contains mobile app translations in a "long" format.
... | TMax-15K task task_003123_578631d1 (data_processing, complex), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003123_578631d1",
"description": "TMax-15K task task_003123_578631d1 (data_processing, complex), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003123_578631d1/tb2 |
terminal-desk/task_003201_68987b2a | You are a localization engineer managing translation strings for a software application. You have two tab-separated values (TSV) files that need to be merged, cleaned, and analyzed to check for UI text expansion issues.
The files are:
1. `/home/user/base_en.tsv` - Contains the base English strings. Columns: `StringID`... | TMax-15K task task_003201_68987b2a (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003201_68987b2a",
"description": "TMax-15K task task_003201_68987b2a (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003201_68987b2a/tb2 |
terminal-desk/task_003240_89972956 | I am a data scientist working on a new project and I need to build a reproducible Bash pipeline to clean some raw sensor datasets.
I have a set of raw CSV files located in `/home/user/raw_data/`. Each file contains sensor readings with the following headers:
`timestamp,sensor_id,temperature,humidity,status`
Unfortun... | TMax-15K task task_003240_89972956 (data_science, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003240_89972956",
"description": "TMax-15K task task_003240_89972956 (data_science, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003240_89972956/tb2 |
terminal-desk/task_003273_5d399c70 | You are a bioinformatics analyst working on a sequence evolution project. Your team uses a custom nonlinear mathematical model to estimate the evolutionary divergence time ($t$) between two genetic sequences based on the observed mutation proportion ($p$).
The model equation is:
$f(t) = 1 - e^{-0.5t} + 0.1t - p = 0$
... | TMax-15K task task_003273_5d399c70 (scientific_computing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003273_5d399c70",
"description": "TMax-15K task task_003273_5d399c70 (scientific_computing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-form... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003273_5d399c70/tb2 |
terminal-desk/task_003350_3d43b53b | You are a data analyst troubleshooting an issue with a company's organizational chart. You requested a dump of the current employee-manager hierarchy from the HR SQLite database. However, due to a corrupted index returning stale rows, the database exported a raw, historical event log instead of the current active snaps... | TMax-15K task task_003350_3d43b53b (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003350_3d43b53b",
"description": "TMax-15K task task_003350_3d43b53b (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003350_3d43b53b/tb2 |
terminal-desk/task_003418_2e98d16a | You are a data scientist cleaning a raw environmental dataset using bash utilities.
You have a dataset located at `/home/user/raw_sensors.csv` in a "wide" format.
The file contains a header and the following comma-separated columns:
`sensor_id,date,metric,v0,v1,v2,v3,v4,v5`
The columns `v0` through `v5` represent re... | TMax-15K task task_003418_2e98d16a (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003418_2e98d16a",
"description": "TMax-15K task task_003418_2e98d16a (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
"ve... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003418_2e98d16a/tb2 |
terminal-desk/task_003457_a10d7c3a | As a backup administrator, you are tasked with processing a legacy data archive. The file is located at `/home/user/archive.zip`.
Your task is to securely extract, sanitize, and convert the contents of this archive following these exact steps:
1. **Nested Archive Extraction**: Extract the `.tar` files contained insid... | TMax-15K task task_003457_a10d7c3a (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003457_a10d7c3a",
"description": "TMax-15K task task_003457_a10d7c3a (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003457_a10d7c3a/tb2 |
terminal-desk/task_003472_edee1bc0 | You are a security researcher analyzing a suspicious data transformation pipeline on a Linux system. You have intercepted a script that a suspected malware strain uses to obfuscate numerical payload data.
In your home directory (`/home/user`), you will find three files:
1. `data_inputs.txt`: A file containing 10,000 f... | TMax-15K task task_003472_edee1bc0 (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003472_edee1bc0",
"description": "TMax-15K task task_003472_edee1bc0 (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifier":... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003472_edee1bc0/tb2 |
terminal-desk/task_003514_5b4b3ea0 | I am a researcher organizing a massive dataset of sensor readings, and I need your help to navigate the nested archives and extract specific anomalous data points using bash commands.
My data is packaged in an archive located at `/home/user/research_data/master_dataset.tar`.
Inside this master archive, there are two ... | TMax-15K task task_003514_5b4b3ea0 (file_operations, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003514_5b4b3ea0",
"description": "TMax-15K task task_003514_5b4b3ea0 (file_operations, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-syst... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003514_5b4b3ea0/tb2 |
terminal-desk/task_003548_45ac0509 | You are a machine learning engineer tasked with preparing a training dataset from a large-scale raw data dump.
You have been provided with a large CSV file located at `/home/user/raw_data.csv`. This file contains 100,000 rows and 10 columns of numerical raw features (floats). Due to the size of typical datasets in ou... | TMax-15K task task_003548_45ac0509 (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003548_45ac0509",
"description": "TMax-15K task task_003548_45ac0509 (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasonin... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003548_45ac0509/tb2 |
terminal-desk/task_003559_2686604b | You are a security researcher analyzing a suspicious binary. You have recovered the source code for a custom decryption tool written in Rust by the malware author, located at `/home/user/malware_decoder`. It is supposed to decrypt a payload located at `/home/user/evidence.enc` and write the result to `/home/user/decryp... | TMax-15K task task_003559_2686604b (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003559_2686604b",
"description": "TMax-15K task task_003559_2686604b (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifier":... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003559_2686604b/tb2 |
terminal-desk/task_003679_b22644c9 | You are an IT support technician responding to an urgent ticket regarding a failed reporting pipeline.
**Ticket Description:**
"Our overnight metric processing job crashed. The script `/home/user/calculate_cooling.awk` parses `/home/user/system_metrics.log` but is failing with a mathematical error (likely a divide-by-... | TMax-15K task task_003679_b22644c9 (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003679_b22644c9",
"description": "TMax-15K task task_003679_b22644c9 (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
},... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003679_b22644c9/tb2 |
terminal-desk/task_003721_a012b394 | You are a machine learning engineer preparing training data. As part of a data cleaning pipeline, you need to filter out covariance matrices that are not numerically stable (i.e., not strictly positive definite) and compute a feature from the stable ones.
You have a dataset of 3x3 symmetric matrices stored in `/home/u... | TMax-15K task task_003721_a012b394 (scientific_computing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003721_a012b394",
"description": "TMax-15K task task_003721_a012b394 (scientific_computing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003721_a012b394/tb2 |
terminal-desk/task_003857_c7cb8d02 | You are a QA engineer setting up a test environment to validate a custom metrics tracking library alongside an expression evaluator.
You have been provided with two files in `/home/user/`:
1. `metrics.c` - A C source file containing a function `void record_metric(double val)` that records metrics.
2. `data.bin` - A b... | TMax-15K task task_003857_c7cb8d02 (software_engineering, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003857_c7cb8d02",
"description": "TMax-15K task task_003857_c7cb8d02 (software_engineering, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003857_c7cb8d02/tb2 |
terminal-desk/task_003917_60ceae0a | You are an analyst auditing a data processing pipeline. A previous analyst made a "data leakage" mistake by normalizing a machine learning test dataset using its own mean and standard deviation, rather than the statistics from the training set.
Your task is to fix this pipeline using ONLY Bash shell built-ins and sta... | TMax-15K task task_003917_60ceae0a (data_science, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003917_60ceae0a",
"description": "TMax-15K task task_003917_60ceae0a (data_science, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reaso... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003917_60ceae0a/tb2 |
terminal-desk/task_003936_1fd895f1 | I am a technical writer organizing documentation for our new product release, and the engineering team sent me a messy, nested archive of draft files. I need you to clean this up and package the final documentation.
Here is what you need to do:
1. Extract the primary archive located at `/home/user/raw_docs.tar.gz`.
2.... | TMax-15K task task_003936_1fd895f1 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003936_1fd895f1",
"description": "TMax-15K task task_003936_1fd895f1 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "media-content-production"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003936_1fd895f1/tb2 |
terminal-desk/task_003949_6966fd6b | You are a performance engineer tasked with optimizing a slow bioinformatics pipeline that analyzes spatial transcriptomics meshes. The pipeline maps molecular sequence differences across a physical mesh grid.
Currently, the analysis is done manually using an inefficient combination of spreadsheets and slow tools. You... | TMax-15K task task_003949_6966fd6b (scientific_computing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003949_6966fd6b",
"description": "TMax-15K task task_003949_6966fd6b (scientific_computing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003949_6966fd6b/tb2 |
terminal-desk/task_003955_d0516932 | You are a data engineer tasked with building a lightweight ETL pipeline using Python.
You have two raw data sources:
1. `/home/user/customers.csv`: Contains customer metadata. Columns: `id`, `age`, `status`.
2. `/home/user/logs.jsonl`: Contains text logs from users. Keys: `user_id`, `text`.
Write and execute a Pytho... | TMax-15K task task_003955_d0516932 (data_science, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003955_d0516932",
"description": "TMax-15K task task_003955_d0516932 (data_science, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"v... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003955_d0516932/tb2 |
terminal-desk/task_003974_2279c3f4 | You are a Database Reliability Engineer (DBRE) managing a complex backup infrastructure. You have exported the metadata of your system's backups into two CSV files located in your home directory:
1. `/home/user/backups.csv`
Columns: `backup_id`, `timestamp`, `size_mb`, `type`
`type` is either `FULL` or `INC` (in... | TMax-15K task task_003974_2279c3f4 (data_querying, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003974_2279c3f4",
"description": "TMax-15K task task_003974_2279c3f4 (data_querying, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003974_2279c3f4/tb2 |
terminal-desk/task_003995_62612a1c | You are an operations engineer triaging an incident where a critical simulation model has started failing to converge after a configuration update.
You have been provided with the source code for the simulation model at `/home/user/sim_model.c` and the failing configuration file at `/home/user/config.txt`.
Your task... | TMax-15K task task_003995_62612a1c (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_003995_62612a1c",
"description": "TMax-15K task task_003995_62612a1c (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
},
"verifier... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_003995_62612a1c/tb2 |
terminal-desk/task_004046_c60f9ea5 | You are a data engineer tasked with building an ETL pipeline step that extracts specific hierarchical metrics from an undocumented SQLite database.
You have been provided with an SQLite database file at `/home/user/company.db`. The schema is completely unknown to you. Your task is to:
1. **Reverse Engineer the Schem... | TMax-15K task task_004046_c60f9ea5 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004046_c60f9ea5",
"description": "TMax-15K task task_004046_c60f9ea5 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004046_c60f9ea5/tb2 |
terminal-desk/task_004063_eed44624 | You are a data analyst tasked with building and testing a Bash-based data processing pipeline. You have been provided with two data feeds: a transactions file and an exchange rate file.
Your goal is to build a robust shell script to join these sources, calculate daily totals in USD with high numerical accuracy, and wr... | TMax-15K task task_004063_eed44624 (data_science, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004063_eed44624",
"description": "TMax-15K task task_004063_eed44624 (data_science, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
"... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004063_eed44624/tb2 |
terminal-desk/task_004108_0d6e2a84 | You are a security researcher analyzing a suspicious Python process that was caught beaconing to a command and control (C2) server.
You have been provided with two files in the `/home/user/forensics/` directory:
1. `memdump.raw` - A raw memory dump of the compromised Python process.
2. `decode.py` - A reverse-enginee... | TMax-15K task task_004108_0d6e2a84 (debugging, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004108_0d6e2a84",
"description": "TMax-15K task task_004108_0d6e2a84 (debugging, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
},
"verifie... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "cybersecurity"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004108_0d6e2a84/tb2 |
terminal-desk/task_004151_675eeec3 | You are a data engineer tasked with building a high-performance mathematical ETL pipeline. We have a batch of unnormalized 10-dimensional vector data that needs to be standardized, processed in parallel, and bulk-loaded into a relational database for a downstream machine learning system.
Here are your instructions:
1... | TMax-15K task task_004151_675eeec3 (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004151_675eeec3",
"description": "TMax-15K task task_004151_675eeec3 (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reaso... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "mathematics-or-formal-reasoning"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004151_675eeec3/tb2 |
terminal-desk/task_004153_e4e8c9ff | You are a data scientist cleaning up and analyzing user feedback logs from two consecutive days.
You have been provided with two files containing user feedback, but they were exported from different systems and are in different formats:
1. `/home/user/day1.csv` - A CSV file. The text you need is in the `feedback` col... | TMax-15K task task_004153_e4e8c9ff (data_processing, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004153_e4e8c9ff",
"description": "TMax-15K task task_004153_e4e8c9ff (data_processing, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004153_e4e8c9ff/tb2 |
terminal-desk/task_004179_a9dd8df5 | You are acting as a backup administrator. A log rotation script on our servers often races with the writing process, leaving some log files with truncated, invalid lines at the end.
We need to archive all fatal error logs from our application servers.
Your task is to:
1. Recursively traverse the directory `/home/use... | TMax-15K task task_004179_a9dd8df5 (file_operations, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004179_a9dd8df5",
"description": "TMax-15K task task_004179_a9dd8df5 (file_operations, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004179_a9dd8df5/tb2 |
terminal-desk/task_004180_1dfac5f4 | We have an interactive bash script at `/home/user/setup_mounts.sh` that our team uses to generate user-space mount configurations. We are migrating our server provisioning to an automated CI/CD pipeline, and this interactive script is blocking our automation.
Your task is to write an `expect` script that automates the... | TMax-15K task task_004180_1dfac5f4 (system_administration, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004180_1dfac5f4",
"description": "TMax-15K task task_004180_1dfac5f4 (system_administration, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-s... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "industrial-physical-systems"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004180_1dfac5f4/tb2 |
terminal-desk/task_004328_99f1edc9 | You are a Database Administrator working on query optimization. The development team has provided you with a NoSQL dump of database access events in JSON Lines format, located at `/home/user/events.jsonl`.
Your task is to write a script (in the language of your choice) that mimics a NoSQL aggregation pipeline to proc... | TMax-15K task task_004328_99f1edc9 (data_querying, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004328_99f1edc9",
"description": "TMax-15K task task_004328_99f1edc9 (data_querying, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
},
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "office-white-collar"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004328_99f1edc9/tb2 |
terminal-desk/task_004361_0d2c270d | You are a support engineer tasked with collecting diagnostics for a sudden reporting failure.
Earlier today, a colleague accidentally deleted a critical script located at `/home/user/report.py`. Fortunately, an instance of this script is still running (hanging or sleeping) in the background.
Your tasks are:
1. **Re... | TMax-15K task task_004361_0d2c270d (debugging, short), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004361_0d2c270d",
"description": "TMax-15K task task_004361_0d2c270d (debugging, short), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
},
"ver... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "software-engineering"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004361_0d2c270d/tb2 |
terminal-desk/task_004386_ee518608 | You are an IT support technician handling an escalated ticket (Ticket #1029).
The billing department reports that their nightly aggregation pipeline is failing. The pipeline is supposed to calculate the total cloud usage costs across all servers by summing the `cost` values in the daily logs. However, the system is c... | TMax-15K task task_004386_ee518608 (debugging, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004386_ee518608",
"description": "TMax-15K task task_004386_ee518608 (debugging, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
},
"ver... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "finance-economics"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004386_ee518608/tb2 |
terminal-desk/task_004444_25557d8a | You are acting as a data scientist for a structural biology lab. We have a protein structure file located at `/home/user/protein.pdb`.
Your task is to estimate the posterior distribution of the temperature factors (B-factors) of the Alpha-Carbon (CA) atoms in this structure.
Please perform the following steps:
1. Pa... | TMax-15K task task_004444_25557d8a (scientific_computing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004444_25557d8a",
"description": "TMax-15K task task_004444_25557d8a (scientific_computing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004444_25557d8a/tb2 |
terminal-desk/task_004470_b3d3970e | You are a performance engineer tasked with optimizing a bioinformatics data processing workflow.
In `/home/user/pipeline`, there is a basic bash-based workflow orchestration script named `orchestrator.sh` that sequentially executes three steps on a dataset of DNA sequences (`/home/user/pipeline/sequences.txt`):
1. `s... | TMax-15K task task_004470_b3d3970e (scientific_computing, moderate), served as a Terminal-Bench 2 task. | [] | 1.1 | {
"schema_version": "1.1",
"task": {
"name": "terminal-desk/task_004470_b3d3970e",
"description": "TMax-15K task task_004470_b3d3970e (scientific_computing, moderate), served as a Terminal-Bench 2 task."
},
"metadata": {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
... | {
"source": "allenai/TMax-15K (ODC-BY)",
"arena_domain": "natural-science"
} | [
"environment/Dockerfile",
"instruction.md",
"task.toml",
"tests/test.sh",
"tests/test_final_state.py"
] | tasks/task_004470_b3d3970e/tb2 |
End of preview. Expand in Data Studio
Terminal Desk
An OpenEnv environment for the OpenEnv Arena. Each episode is one
command-line task in a fresh Ubuntu 22.04 container. The model works in a shell (exec, write_file). When it calls
evaluate, or the episode ends, a pytest verifier checks the final state of the filesystem.
What it teaches
Multi-step terminal work across the Arena's eight domains. The tasks cover software engineering, industrial and physical systems, natural science, office work, finance and economics, mathematics, cybersecurity, and media/content production. The model has to inspect files, write scripts and transform data, then leave the system in the state the instruction asks for. All of this happens within a 32-tool-call, 4,096-completion-token budget.
Interface
Tasks are served through OpenEnv's tbench2_env (vendored unchanged in base/tbench2_env, pinned to OpenEnv
86a180e), configured the way the Arena operator's Terminal-Bench images are:
- local mode;
- tests withheld from the agent;
- 110 s command timeout.
Each task is its own image: one thin layer over a shared base.
Raw source
Tasks come from allenai/TMax-15K (ODC-BY), and this dataset keeps
that license. Each task has a TMax container.def setup, a task description and a test_final_state.py verifier.
convert.py turns these into Terminal-Bench 2 task directories without changing the description or the verifier.
TMax-15K has no overlap with Terminal-Bench 2, and no Terminal-Bench 2 task is used.
How tasks were chosen
- Candidates (
select_candidates.py,label_domains.py). Claude Haiku 5.5 labelled each TMax task with the Arena domain it most exercises. Tasks needing packages outside the base image's allowlist were dropped. That left 160 candidates, 20 per domain. - Build check. Every task image is built. A task is kept only if its verifier scores exactly 0 on the untouched initial state.
- Mining.
- The base model, Qwen3.8-27B (thinking off, temperature 0.7), ran 8 rollouts per task. Replies were parsed leniently: the first JSON object counts.
- A task is Learnable when the base model solves 2–6 of 8, and a reference model solves at least 3 of 4. The reference is Qwen3.8-2.4T-A95B at low reasoning effort, or Claude Haiku 5.5. The reference's completion budget excludes its reasoning tokens: the reference only shows a task can be solved, and the base model's band, under the real budget, sets difficulty.
- Selection. 50 Learnable tasks, balanced across the eight domains (6–7 each), preferring tasks the base model solves about half the time.
Reward
Binary. 1.0 if every test in the task's test_final_state.py passes on the final container state, else 0.0.
Files
tasks.jsonl: the 50 submitted tasks (task_id,arena_domain,tmax_domain,skill_type,complexity,tier).tasks/<task_id>/: one directory for each of the 160 candidates.Dockerfile,post.sh,files/: the task image.tb2/: the Terminal-Bench 2 task (instruction, verifier,task.toml).
selection.json,mining.json: the selected ids and the mining results for each task.base/: the shared base image and the vendoredtbench2_env.select_candidates.py,label_domains.py,convert.py,candidates.json: the generator.
- Downloads last month
- 119