earakely-scale/VaultEnv-REA

vault

vault

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

Tables

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

Run it

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

HashiCorp Vault on AgentEnv, from a REA capture

What this is. A replay environment of HashiCorp Vault 1.21.4 (BUSL-1.1), built from one recorded browser session against a self-hosted instance seeded with synthetic data. REA projected the capture into an environment manifest (43 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 HashiCorp Vault. Tasks are graded deterministically, with no judge.

What's in it

• 43 tools (36 GET, 3 POST, 3 DELETE, 1 PUT) over 85 recorded exchanges on http://127.0.0.1:3303; no recorded group excluded.
• Path parameters: 5 uuid, 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 44 tools.
• Unknowns the manifest states explicitly: authorization model, response semantics beyond recorded exchanges, state transitions between tools.
• 12 retrieval tasks (evals/vault-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_0f0dc760f8613dc5c13f30aa0c5199dd83579261e8233758d9beb3e2c0ea8c87 (scoped HAR sha256 02038b8412236a14ef63eaa9155a49ab89c2ad927d2c2d9fb2ab73b39430e6ea, 176,735 bytes), manifest view ev_f813cf623fb7f6c6698adbe68d07fd4ce70f881b4802af308d4a62eac949826d, 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_identity_group_id_by_uuid	get_identity_group_id_by_uuid	uuid=5c2aa432-e000-7add-98f4-cd8cfafb5c46	data.metadata.purpose	Quarry audits and stone counts
qa-02-get_identity_entity_alias_id_by_uuid	get_identity_entity_alias_id_by_uuid	uuid=99156b6f-351c-184c-e10d-6c5322b5defe	data.name	ada.finch
qa-03-get_identity_entity_id_by_uuid	get_identity_entity_id_by_uuid	uuid=f220efed-73df-c808-83a6-35c2b2fd0647	data.metadata.display_name	Ada Finch
qa-04-get_sys_internal_ui_mounts_rota_config	get_sys_internal_ui_mounts_rota_config	no arguments	data.description	Shift and station settings for the Lighthouse Rota
qa-05-get_sys_auth	get_sys_auth	no arguments	approle/.description	Machine sign-in reserved for the Saltmarsh Depot scanners
qa-06-get_rota_config_data_saltmarsh_depot_inv	get_rota_config_data_saltmarsh_depot_inventory	no arguments	data.data.tier	bulk storage tier
qa-07-get_sys_mounts	get_sys_mounts	no arguments	identity/.description	identity store
qa-08-get_rota_config_data_blue_heron_kitchen_	get_rota_config_data_blue_heron_kitchen_menu_cycle	no arguments	data.data.tier	galley standard
qa-09-get_ledger_config_metadata_granite_ledge	get_ledger_config_metadata_granite_ledger_quarry_audit	no arguments	data.custom_metadata.steward	Rui Castellanos
qa-10-get_rota_config_metadata_saltmarsh_depot	get_rota_config_metadata_saltmarsh_depot_inventory	no arguments	data.custom_metadata.review_board	Saltmarsh Depot Yard Committee
qa-11-get_rota_config_data_lighthouse_rota_shi	get_rota_config_data_lighthouse_rota_shift_plan	no arguments	data.data.handover_desk	Tide Table Desk
qa-12-get_ledger_config_data_granite_ledger_qu	get_ledger_config_data_granite_ledger_quarry_audit	no arguments	data.data.block_count	412 blocks north quarry

Episodes

model	tasks	passed	pass rate	mean tool calls	mean input tokens
openai/gpt-oss-120b:groq	12	11	92%	0.9	10,107
zai-org/GLM-5.3-Flash:zai-org	12	12	100%	1.0	12,375

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

openai/gpt-oss-120b:groq missed qa-11-get_rota_config_data_lighthouse_rota_shi. Miss by gpt-oss-120b:groq on qa-11-get_rota_config_data_lighthouse_rota_shi: the model provider rejected a malformed tool call the model generated (HTTP 400), so the episode ended without an answer.

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/VaultEnv-REA --bundle vault --task smoke          # Docker, no model
agent-env hf run earakely-scale/VaultEnv-REA --bundle vault --eval vault-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. HashiCorp Vault ran in Docker on a throwaway VM, bound to loopback, seeded through its own API with synthetic records (kv v2 mounts: rota-config: Shift and station settings for the Lighthouse Rota; ledger-config: Quarry audit settings kept by the Granite Ledger crew; kv entries: rota-config/lighthouse-rota/shift-plan, rota-config/lighthouse-rota/tide-tables, rota-config/blue-heron-kitchen/menu-cycle, rota-config/saltmarsh-depot/inventory, rota-config/copperfield-annex/archive, ledger-config/granite-ledger/quarry-audit; kv entry keys: owner, region, rotation_days, tier plus one or two configuration fields each (handover_desk, covers_per_sitting, bay_count, shelf_code, block_count, ...); four entries carry...); 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	HashiCorp Vault (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)	43 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 HashiCorp Vault; 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 HashiCorp Vault instance over synthetic data; HashiCorp Vault is BUSL-1.1 software by HashiCorp, Inc. 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.