earakely-scale/ImmichEnv-REA

immich

immich

An AgentEnv bundle with its recorded runs, published with agentenv-hf.

Tables

• immich_tasks: one row per task of the bundle, with its prompt, envs, verifiers and steps.
• immich_episodes: one row per recorded run, with its status, model, reward and scores, the prompt and response, and the agent's transcript as chat messages when its trajectory format is one agentenv-hf reads.

raw/immich.jsonl holds each run's record and native trajectory, and bundles/immich/ the bundle itself.

Run it

pip install agentenv-framework huggingface_hub
hf download earakely-scale/ImmichEnv-REA --repo-type dataset --local-dir agentenv-dataset
agent-env run agentenv-dataset/bundles/immich

Immich on AgentEnv, from a REA capture

What this is. A replay environment of Immich 3.3.3 (AGPL-3.0), built from one recorded browser session against a self-hosted instance seeded with synthetic data. REA projected the capture into an environment manifest (45 tools), and rea environment-export wrote the bundle you can run from this dataset. Every tool call replays a retained record from the capture, cites the record and the Evidence it came from, and nothing executes Immich. Tasks are graded deterministically, with no judge.

What's in it

• 45 tools (41 GET, 2 POST, 1 DELETE, 1 PATCH) over 120 recorded exchanges on http://127.0.0.1:3207; no recorded group excluded.
• Path parameters: 13 uuid. REA's four format rules (numeric, uuid, hex, iso-date) applied unconditionally; the three opt-in rules (after-uid-literal, opaque-token, sibling-variance) were requested and are labelled per parameter. With the four format rules alone the manifest has 45 tools.
• Unknowns the manifest states explicitly: authorization model, response semantics beyond recorded exchanges, state transitions between tools.
• 12 retrieval tasks (evals/immich-retrieval.toml): each asks one factual question about a seeded object, with a distinct answer that the app does not ship by default, answerable through a recorded exchange; reward 1 if the response contains the recorded value, else 0. Plus smoke, which deploys the environment and checks replay through the MCP endpoint.
• 24 recorded episodes by openai/gpt-oss-120b:groq, zai-org/GLM-5.3-Flash:zai-org through the Hugging Face router, with their chat transcripts.
• Provenance: capture ev_1a9cf44e9f31412cd6f46b77293c666f4bfa1f6f9afccc430ca3b8d83937590a (scoped HAR sha256 a525d5143fb73344d07d44ce129c5d9be7d83ef9a2d5dfc2f453bfa80b0647d0, 228,076 bytes), manifest view ev_fda1218791c0c5ced99ef79739463e0a102755b2cdf71df008e93bd52e3ba617, confidence inferred, authority historical-reference. The REA build is a prerelease of the environment commands proposed upstream in morluto/rea#4954, #4955, #4975 and #4999; the bundle installs it from the tarball it ships, so the environment runs without waiting for an npm release.

What replay means, and does not

A tool call returns the recorded response whose path and query values best match the arguments, with replay.exact_match telling you whether it matched exactly or fell back to the closest record. Reads of seeded objects behave like the application. Writes return the recorded answer of the one write the session made, and the environment has no state: a created object does not appear in a later list. This is what makes retrieval tasks hard-gradable and write tasks weak; the manifest's unknowns say so.

Tasks and grading

task	tool	arguments	field	answer
qa-01-get_shared_links_by_uuid	get_shared_links_by_uuid	uuid=23ef136d-1419-42bd-9d2a-b3556e6a1a99	description	Night shift photos for the relief crew at Granite Point
qa-02-get_admin_users_by_uuid	get_admin_users_by_uuid	uuid=c5feb885-5c35-43a2-840b-c20b64eca2b5	storageLabel	ada-finch
qa-03-get_tags_by_uuid	get_tags_by_uuid	uuid=a24f3278-8a9c-4375-9f65-b24a5ccdc6c3	name	Saltmarsh Depot
qa-04-get_people_by_uuid	get_people_by_uuid	uuid=49d85fa6-9c22-49b2-bdc6-5617be17efbd	name	Mina Okafor
qa-05-get_assets_by_uuid	get_assets_by_uuid	uuid=87406a9b-ff65-45a0-b65a-fe59cbfc04c9	originalFileName	granite-ledger-north-quarry.png
qa-06-get_libraries_by_uuid	get_libraries_by_uuid	uuid=eb0affe1-df42-4edf-b96d-621be4f516a1	name	Saltmarsh Depot Archive
qa-07-get_albums_by_uuid	get_albums_by_uuid	uuid=78219be3-6308-4ea3-a8ed-2a60633845bb	description	Photos from the winter night shift, three keepers on a 22:00 to 06:00 rota.
qa-08-get_notifications	get_notifications	unread=true	[0].title	Granite Ledger audit due
qa-09-get_albums	get_albums	no arguments	[0].description	Tide-table menus and plating from the seaward kitchen at Granite Point.
qa-10-get_shared_links	get_shared_links	no arguments	[0].description	Stairwell repaint proof for the Copperfield Annex committee
qa-11-get_tags	get_tags	no arguments	[0].name	Copperfield Annex
qa-12-get_admin_users	get_admin_users	withDeleted=true	[0].email	rui.castellanos@example.test

Episodes

model	tasks	passed	pass rate	mean tool calls	mean input tokens
openai/gpt-oss-120b:groq	12	12	100%	1.0	11,492
zai-org/GLM-5.3-Flash:zai-org	12	12	100%	1.1	15,078

Every episode is in the episodes config with its transcript; rewards are the deterministic verifier scores.

Every recorded episode passed.

Run it yourself

uv tool install agentenv-framework --with 'agentenv-hf @ git+https://github.com/earakely-scale/agentenv-hf-plugin@v0.3.0'
agent-env hf run earakely-scale/ImmichEnv-REA --bundle immich --task smoke          # Docker, no model
agent-env hf run earakely-scale/ImmichEnv-REA --bundle immich --eval immich-retrieval --model openai/gpt-oss-120b:groq

The retrieval eval needs a model endpoint: set [model] base_url and api_key in .agentenv/config.toml (the Hugging Face router works with any hf_ token). The bundle's agents/replay-qa is a small OpenAI-compatible tool-calling agent; any A2A agent can replace it.

How it was made

1. Immich ran in Docker on a throwaway VM, bound to loopback, seeded through its own API with synthetic records (admin Rea Admin plus users Ada Finch and Rui Castellanos; 5 generated PNG photos with descriptions, ratings and favourites; 3 albums (Lighthouse Rota Night Shift, Granite Ledger Field Notes, Blue Heron Kitchen Menus), two shared with users; 4 tags (Saltmarsh Depot with nested Cart Inventory, Copperfield Annex, Tide Tables); 2 people (Mina Okafor, Tomas Vey) on manual faces; 2 shared links; 1 album comment; 1 partner share; 1 external library (Saltmarsh Depot Archive); 1 notification); people and projects are invented.
2. Playwright drove a login, the main pages and a scripted read, write, read sequence through the app's JSON API; the HAR was scoped to the API, the login exchange dropped, secret-looking body fields scrubbed, headers and cookies removed.
3. rea inspect-web-network-capture retained the exchanges as Evidence; rea inspect-analysis-view projected the environment-manifest view; rea environment-export wrote this bundle; gen_tasks.py derived the retrieval tasks from the manifest and the recorded bodies; agent-env run recorded the episodes; agent-env hf publish wrote this dataset. Scripts: https://github.com/earakely-scale/rea-environments.

Compared with PortSimEnv on AgentEnv

	PortSimEnv-AgentEnv	Immich (this dataset)
Environment	A simulation: a container quay, ships, cranes, weather, with A2A ship agents	A replay of a real application's API, captured from one session
Observation	Tool results computed by the simulator as the week unfolds	Recorded responses, cited to Evidence
Action space	Planning tools (check_plan, confirm_berths, advance, message_agent)	45 tools mirroring the application's endpoints
State	Full: a virtual clock, triggers, frozen windows, agents that answer	None: every call is independent; writes replay one recorded answer
Reward	Deterministic, against a CP-SAT optimum or a re-planner anchor; continuous 0 to 1	Deterministic, recorded-value match; 0 or 1 per task
What it measures	Planning under uncertainty and negotiation	Tool discovery, parameter inference and faithful retrieval in a real API
Data provenance	Real 2024 port calls and weather; simulated workloads	Synthetic seed data on a self-hosted Immich; nothing real
How it was built	Hand-written plugin, months of work	Generated from a capture in one scripted pass
Where it is strong	Long-horizon decisions with hard rewards	Breadth: any app you can capture becomes an environment

The two are complementary: PortSimEnv shows what a hand-built simulation can grade; this dataset shows how far a capture alone gets you, and the manifest's unknowns list exactly what a stateful version would still have to specify.

Licence and credits

Bundle code and tasks: Apache-2.0. Recorded content: responses generated by a self-hosted Immich instance over synthetic data; Immich is AGPL-3.0 software by the Immich contributors and is not redistributed here beyond the recorded JSON responses. REA is MIT software by morluto; AgentEnv Framework is Apache-2.0 by Scale AI.