earakely-scale/HealthchecksEnv-REA

healthchecks

healthchecks

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

Tables

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

Run it

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

Healthchecks on AgentEnv, from a REA capture

What this is. A replay environment of Healthchecks 4.4 (BSD-3-Clause), built from one recorded browser session against a self-hosted instance seeded with synthetic data. REA projected the capture into an environment manifest (18 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 Healthchecks. Tasks are graded deterministically, with no judge.

What's in it

• 18 tools (13 GET, 4 POST, 1 DELETE) over 69 recorded exchanges on http://127.0.0.1:3501; no recorded group excluded.
• Path parameters: 10 uuid, 2 hex, 2 numeric, 2 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 20 tools.
• Unknowns the manifest states explicitly: authorization model, response semantics beyond recorded exchanges, state transitions between tools.
• 12 retrieval tasks (evals/healthchecks-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_a20c9a68a2e024e0f536f64dbc25e033345d026c5d1e2bd9293918666ed8ec18 (scoped HAR sha256 86c4ddd7fe0414a97b35f2ca0d41a802c7ca60fd1c2109006d1661926a21e601, 156,615 bytes), manifest view ev_0c2233edd0b12179b475e94622d4b57d5c9402a1d46ace906f4a49cc8f7de3a6, 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_badge_fa5b0485_8fec_41ca_b4bf_9b870c	get_badge_fa5b0485_8fec_41ca_b4bf_9b870c_by_fa5b0485_8fec_41	fa5b0485_8fec_41ca_b4bf_9b870c_id=Ee3sT1UG-2.shields	label	Saltmarsh Depot Board
qa-02-get_checks_status_by_uuid	get_checks_status_by_uuid	uuid=533dd859-9732-44ed-892b-b911079d760b	title	Granite Ledger nightly backup – Lighthouse Rota Monitor
qa-03-get_checks_pings_by_uuid	get_checks_pings_by_uuid	uuid=56b4d521-e752-4280-93f2-b15934d5edf1	pings[0].ua	BlueHeron-FridgeProbe/1.8 walk-in-3
qa-04-get_projects_checks_status_by_uuid	get_projects_checks_status_by_uuid	uuid=ab43b910-eeb3-400a-92f3-fd435a7a3f82	title	1 down – Lighthouse Rota Monitor
qa-05-get_checks_by_uuid	get_checks_by_uuid	uuid=533dd859-9732-44ed-892b-b911079d760b	name	Granite Ledger nightly backup
qa-06-get_checks_by_hash	get_checks_by_hash	hash=cb6c10db431dd2738cea646da1d2e0060cf325b0	slug	saltmarsh-depot-tide-gauge-upload
qa-07-get_b_by_id_id_2	get_b_by_id_id_2	id=2, id_2=b45272fe-11c9-4c9e-9fac-18f66a0e9a43.shields	label	Granite Ledger quarry audit export
qa-08-get_checks	get_checks	no arguments	checks[0].slug	granite-ledger-nightly-backup
qa-09-get_channels	get_channels	no arguments	channels[0].name	Saltmarsh Depot on-call email
qa-10-get_checks_by_uuid	get_checks_by_uuid	uuid=533dd859-9732-44ed-892b-b911079d760b	tags	granite-ledger backup nightly
qa-11-get_checks_by_hash	get_checks_by_hash	hash=cb6c10db431dd2738cea646da1d2e0060cf325b0	name	Saltmarsh Depot tide gauge upload
qa-12-get_checks	get_checks	no arguments	checks[0].desc	Dumps the quarry tally database to the Saltmarsh Depot vault at 02:15.

Episodes

model	tasks	passed	pass rate	mean tool calls	mean input tokens
openai/gpt-oss-120b:groq	12	11	92%	1.0	9,004
zai-org/GLM-5.3-Flash:zai-org	12	12	100%	1.0	10,668

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

openai/gpt-oss-120b:groq missed qa-10-get_checks_by_uuid. Miss by gpt-oss-120b:groq on qa-10-get_checks_by_uuid: 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/HealthchecksEnv-REA --bundle healthchecks --task smoke          # Docker, no model
agent-env hf run earakely-scale/HealthchecksEnv-REA --bundle healthchecks --eval healthchecks-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. Healthchecks ran in Docker on a throwaway VM, bound to loopback, seeded through its own API with synthetic records (project: Lighthouse Rota (owner rea-admin@example.test, read-write and read-only API keys minted by seed.py, kept only in .admin); checks: Granite Ledger nightly backup (cron 15 2 * * * UTC, up, start + success pings), Blue Heron Kitchen fridge sensor (timeout 600 s, POST-only, 3 pings), Copperfield Annex certificate renewal (cron 0 6 * * 1, down after a /fail ping), Saltmarsh Depot tide gauge upload (timeout 900 s, exit-status 0 ping), Lighthouse Rota lamp oil report (cron 30 7 * * 5, manual resume, paused), Granite Ledger quarry audit export (timeout 1 day, keyword filters...); 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	Healthchecks (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)	18 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 Healthchecks; 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 Healthchecks instance over synthetic data; Healthchecks is BSD-3-Clause software by Pēteris Caune and the Healthchecks 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.