earakely-scale/PocketBaseEnv-REA

pocketbase

pocketbase

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

Tables

• pocketbase_tasks: one row per task of the bundle, with its prompt, envs, verifiers and steps.
• pocketbase_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/pocketbase.jsonl holds each run's record and native trajectory, and bundles/pocketbase/ the bundle itself.

Run it

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

PocketBase on AgentEnv, from a REA capture

What this is. A replay environment of PocketBase 0.40.5 (MIT), built from one recorded browser session against a self-hosted instance seeded with synthetic data. REA projected the capture into an environment manifest (25 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 PocketBase. Tasks are graded deterministically, with no judge.

What's in it

• 25 tools (21 GET, 2 POST, 1 DELETE, 1 PATCH) over 86 recorded exchanges on http://127.0.0.1:3506; no recorded group excluded.
• Path parameters: 7 opaque-token. 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 38 tools.
• Unknowns the manifest states explicitly: authorization model, response semantics beyond recorded exchanges, state transitions between tools.
• 12 retrieval tasks (evals/pocketbase-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_54a03a4c303ea4124958962eb7804b995de94c2a8d2470c380a037599f8827b1 (scoped HAR sha256 66f2912459356dfbfeecee9496896dea963018cc482b561c5bd4b6b754e1e042, 631,052 bytes), manifest view ev_3de14d4e339bfe06f7aa6ae910aea22c80448002da9ff04ac5413b546032301d, confidence inferred, authority historical-reference. The REA build is rea-agents 6.7.0 from main at commit 3fca2e0c, which merged the environment commands (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 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_collections_crew_records_by_uid	get_collections_crew_records_by_uid	uid=znw0je8t1y5kppa	full_name	Mina Okafor
qa-02-get_collections_inspections_records_by_u	get_collections_inspections_records_by_uid	uid=bx0hjvhv00mub79	description	Bonding cable frayed at the clamp; transfers stopped until it is replaced.
qa-03-get_collections_voyages_records_by_uid	get_collections_voyages_records_by_uid	uid=9oeudt67jj74zfu	summary	Rui Castellanos tallied every block at the slip; two cracked blocks went back.
qa-04-get_collections_vessels_records_by_uid	get_collections_vessels_records_by_uid	uid=os6ury77csjddb2	description	Shallow-draught launch; new propeller shaft fitted at Copperfield Annex.
qa-05-get_collections_ports_records_by_uid	get_collections_ports_records_by_uid	uid=h2dmwegrzc73shx	name	Copperfield Annex Wharf
qa-06-get_settings	get_settings	no arguments	meta.senderAddress	registry@example.test
qa-07-get_collections_voyages_records	get_collections_voyages_records	perPage=50, sort=created	items[0].title	Kelp Run to Blue Heron Kitchen
qa-08-get_collections_ports_records	get_collections_ports_records	perPage=50, sort=created	items[0].description	Tidal quay; berths dry out two hours either side of low water.
qa-09-get_collections_crew_records	get_collections_crew_records	perPage=50, sort=created	items[0].bio	Keeps the tide tables for the winter rota and logs every crossing in pencil.
qa-10-get_collections_vessels_records	get_collections_vessels_records	perPage=50, sort=created	items[0].name	Blue Heron
qa-11-get_collections_inspections_records	get_collections_inspections_records	perPage=50, sort=created	items[0].description	Shaft seal weeping slightly; recheck after ten running hours.
qa-12-get_collections_crew_records_by_uid	get_collections_crew_records_by_uid	uid=znw0je8t1y5kppa	bio	Rebuilt the Lantern's port diesel twice and hums sea shanties while she works.

Episodes

model	tasks	passed	pass rate	mean tool calls	mean input tokens
openai/gpt-oss-120b:groq	12	12	100%	1.0	9,498
zai-org/GLM-5.3-Flash:zai-org	12	12	100%	1.0	11,790

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/PocketBaseEnv-REA --bundle pocketbase --task smoke          # Docker, no model
agent-env hf run earakely-scale/PocketBaseEnv-REA --bundle pocketbase --eval pocketbase-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. PocketBase ran in Docker on a throwaway VM, bound to loopback, seeded through its own API with synthetic records (settings: app name Saltmarsh Fleet Registry, sender registry@example.test; collections: ports, vessels (home_port -> ports), crew (vessel -> vessels), voyages (vessel, origin, destination), inspections (vessel, inspector -> crew); superuser-only rules; ports: Saltmarsh Depot Quay, Granite Point Landing, Copperfield Annex Wharf, Lighthouse Rota Jetty; vessels: Blue Heron (BHK-114), Granite Ledger Barge (GLD-207), Saltmarsh Wren (SMW-031, refit), Copperfield Lantern (CFL-588), Tidewater Gannet (TWG-660, laid up); crew: Ada Finch, Rui Castellanos, Mina Okafor, Tomas Vey, Petra Quill, Jonah...); 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	PocketBase (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)	25 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 PocketBase; 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 PocketBase instance over synthetic data; PocketBase is MIT software by the PocketBase authors 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.