earakely-scale/ArticCollectionEnv-REA

artic-collection

artic-collection

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

Tables

• artic-collection_tasks: one row per task of the bundle, with its prompt, envs, verifiers and steps.
• artic-collection_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/artic-collection.jsonl holds each run's record and native trajectory, and bundles/artic-collection/ the bundle itself.

Run it

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

Art Institute of Chicago API on AgentEnv, from a REA capture

What this is. A replay environment of Art Institute of Chicago API v1 (CC0 1.0 (collection data)), built from one recorded browser session against the public API at https://api.artic.edu on 2026-10-19. The responses are real public-domain records published by the Art Institute of Chicago, frozen as they were served then. REA projected the capture into an environment manifest (11 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 Art Institute of Chicago API. Tasks are graded deterministically, with no judge.

What's in it

• 11 tools (11 GET) over 26 recorded exchanges on https://api.artic.edu; no recorded group excluded.
• Path parameters: 5 numeric, 1 sibling-variance. 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 12 tools.
• Unknowns the manifest states explicitly: authorization model, response semantics beyond recorded exchanges, state transitions between tools.
• 12 retrieval tasks (evals/artic-collection-retrieval.toml): each asks one factual question about a recorded record, with a distinct answer, 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_5f3f80e48d8e884648a21246b95ac6d60d744e11a841599347d8c8d0bf2a6122 (a capture_browser_scenario session REA recorded itself; capture file sha256 3a807e85c0db318a3cbecbc62f8c7125f4163f6b9c575b6f57f3d2057de5b0a0, 183,942 bytes), manifest view ev_2fb61dbbfdb2a429b27771afd068d7d3d9e0d9d9595da2c62086dfcbefd4b2f2, confidence inferred, authority controlled-replay. The REA build is a prerelease of the environment commands proposed upstream in morluto/rea#4954, #4955, #4975, #4999 and #5182; 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 behave like the API did on 2026-10-19. The live API keeps changing, so every answer is about the recorded snapshot, not today, and a query the session never made falls back to the closest record. The API is read-only and so is the environment: it has no state and no write tools. The manifest's unknowns say what replay cannot know.

Tasks and grading

task	tool	arguments	field	answer
qa-01-get_exhibitions_by_id	get_exhibitions_by_id	id=9036, fields=id,title,status,aic_start_at,aic_end_at,gallery_title,artwork_ids,artist_ids	data.title	Monet and Chicago
qa-02-get_artists_by_id	get_artists_by_id	id=40610, fields=id,title,sort_title,alt_titles,birth_date,death_date,is_artist	data.title	Vincent van Gogh
qa-03-get_artworks_by_id	get_artworks_by_id	id=27992, fields=id,title,artist_display,artist_title,artist_id,date_display,date_start,date_end,medium_display,dimensions,place_of_origin,style_title,classification_title,department_title,gallery_title,gallery_id,artwork_type_title,artwork_type_id,main_reference_number,color,is_public_domain	data.title	A Sunday on La Grande Jatte — 1884
qa-04-get_galleries_by_id	get_galleries_by_id	id=2147478672	data.title	Gallery 240
qa-05-get_category_terms_by_category_term_id	get_category_terms_by_category_term_id	category_term_id=PC-10	data.title	Painting and Sculpture of Europe
qa-06-get_artworks_search	get_artworks_search	q=amphora, limit=5, fields=id,title,date_display,department_title,is_public_domain	data[0].title	Belly-Amphora (Storage Jar)
qa-07-get_artists_search	get_artists_search	q=hokusai, limit=3, fields=id,title,birth_date,death_date	data[0].title	Katsushika Hokusai
qa-08-get_departments	get_departments	limit=20, fields=id,title,subtype,parent_id	data[0].title	Arts of Greece, Rome, and Byzantium
qa-09-get_artworks	get_artworks	ids=27992,28560,16568,24645, fields=id,title,artist_title,date_display	data[0].title	The Bedroom
qa-10-get_exhibitions_by_id	get_exhibitions_by_id	id=9036, fields=id,title,status,aic_start_at,aic_end_at,gallery_title,artwork_ids,artist_ids	data.gallery_title	Gallery 270
qa-11-get_artists_by_id	get_artists_by_id	id=40610, fields=id,title,sort_title,alt_titles,birth_date,death_date,is_artist	data.sort_title	Gogh, Vincent van
qa-12-get_artworks_by_id	get_artworks_by_id	id=27992, fields=id,title,artist_display,artist_title,artist_id,date_display,date_start,date_end,medium_display,dimensions,place_of_origin,style_title,classification_title,department_title,gallery_title,gallery_id,artwork_type_title,artwork_type_id,main_reference_number,color,is_public_domain	data.artist_title	Georges Seurat

Episodes

model	tasks	passed	pass rate	mean tool calls	mean input tokens
openai/gpt-oss-120b:groq	12	11	92%	1.0	3,800
zai-org/GLM-5.3-Flash:zai-org	12	12	100%	1.0	4,695

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

openai/gpt-oss-120b:groq missed qa-06-get_artworks_search. Miss by gpt-oss-120b:groq on qa-06-get_artworks_search: the model's answer did not contain the expected value.

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/ArticCollectionEnv-REA --bundle artic-collection --task smoke          # Docker, no model
agent-env hf run earakely-scale/ArticCollectionEnv-REA --bundle artic-collection --eval artic-collection-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. No instance was run: a throwaway VM opened the public API's URLs in a browser REA launched, within the publisher's fair-access rules (no key, no login; 26 navigations in one browser session with a 600 ms pause after each (under 2 requests per second and under the API's 60 requests per minute); every request passes fields= so each body is 0.6-1.9 KB).
2. REA's capture-browser-scenario visited each API URL (goto) and recorded every network transaction (Real public data: the department list, two artwork searches, an artist search, an exhibition search, a multi-id artwork list, and by-id reads of 10 public-domain artworks across 6 departments, 3 artists, 2 galleries, 2 artwork types, 2 category terms and 1 exhibition.); the environment-manifest view read that capture directly, with no HAR or scrubbing step.
3. 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	Art Institute of Chicago API (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)	11 tools mirroring the application's endpoints
State	Full: a virtual clock, triggers, frozen windows, agents that answer	None: every call is independent; the API is read-only
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	Real public-domain records from the Art Institute of Chicago, as served on 2026-10-19
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 of the public API at https://api.artic.edu, published by the Art Institute of Chicago (CC0 1.0 (collection data)); the dataset repeats them as recorded. REA is MIT software by morluto; AgentEnv Framework is Apache-2.0 by Scale AI.