Workspace-Bench Environment Agent API

An independent demo and callable API for the Workspace-Bench Environment Agent, which drives dynamic workspace evolution from user/environment profiles, workspace state, and historical tasks.

Current generation status:

• The API is framework/backend-driven: EnvironmentAgent diagnoses the workspace, calls a pluggable planning backend, and optionally applies the returned plan to mock workspace state.
• The default backend is mock_rule_based. It uses deterministic hand-written templates for demos, tests, and integration debugging.
• Real non-rule generation should be connected by implementing a PlanningBackend with generation_mode="llm" or generation_mode="external_provider".
• Backend output is validated before workspace evolution. Invalid event-plan-task-workunit references return a structured 422 plan_validation_failed response.
• Unknown mock-store resource IDs return a structured 404 store_resource_not_found response instead of an internal server error.
• Store-backed calls accept an optional run_id so demo simulations can isolate independent runs instead of sharing one mutable mock store. Non-default runs are process-local demo state, capped at 32 active runs, and run_id must match [A-Za-z0-9_.-]{1,48}.

Links

• GitHub repository: https://github.com/outlier27-cell/workspace-bench-environment-agent-api
• Static GitHub Pages demo: https://outlier27-cell.github.io/workspace-bench-environment-agent-api/
• API documentation: API.md

GitHub Pages hosts the visual demo only. It does not run Python services. To expose the callable API publicly, deploy this same repository to a Python-capable host such as Vercel or Render.

Core Idea

persona + environment -> events -> WorkUnits -> artifact ecology -> dependency-aware tasks

The static page shows structured external event sampling, WorkUnit lifecycle changes, workspace snapshots, artifact state changes, dependency graph construction, and task/rubric generation.

The Python API exposes the same Environment Agent flow as callable HTTP endpoints.

Public API Usage

After deploying the repository to Vercel or another Python host, use the deployment domain directly over HTTPS. No local port is needed.

https://<your-api-domain>/api/health
https://<your-api-domain>/api/docs
https://<your-api-domain>/api/environment-agent/simulate

Example request:

curl -X POST "https://<your-api-domain>/api/environment-agent/simulate" \
  -H "Content-Type: application/json" \
  -d '{"user_id":"user_logistics_001","environment_id":"env_peak_logistics","workspace_id":"workspace_logistics_demo","run_id":"demo_run","seed":7,"steps":3,"reset_before_run":true}'

Expected high-level result:

{
  "workspace_id": "workspace_logistics_demo",
  "event_ids": [
    "evt_workspace_logistics_demo_incident_0001_0007",
    "evt_workspace_logistics_demo_policy_change_0002_0008",
    "evt_workspace_logistics_demo_collaboration_0003_0009"
  ],
  "final_workspace_state": {
    "current_snapshot_id": "snap_0004"
  }
}

Main Endpoints

GET  /api/health
GET  /api/profiles/user/{user_id}
GET  /api/profiles/environment/{environment_id}
GET  /api/workspace/{workspace_id}/state
GET  /api/workspace/{workspace_id}/tasks/history
GET  /api/workspace/{workspace_id}/events
GET  /api/workspace/{workspace_id}/inspect
GET  /api/workspace/{workspace_id}/quality
POST /api/workspace/{workspace_id}/reset
GET  /api/environment-agent/backend
POST /api/environment-agent/diagnose
POST /api/environment-agent/step/from-store
POST /api/environment-agent/simulate
POST /api/environment-agent/manager-payload/from-store

Deployment

Static demo:

• GitHub Pages deploys from .github/workflows/pages.yml.
• The static page is served at https://outlier27-cell.github.io/workspace-bench-environment-agent-api/.

Callable API:

• Deploy the same repository to Vercel, Render, or another Python-capable host.
• Vercel uses api/index.py and vercel.json.
• Public API routes are served under /api/....
• Use run_id on store-backed endpoints and workspace query endpoints when you want isolated demo state. Omitting it uses the shared default run. Non-default run_id values are short safe identifiers ([A-Za-z0-9_.-]{1,48}); the API keeps at most 32 active process-local demo runs.

Files

• index.html: page structure
• styles.css: visual design
• app.js: scenario data and interaction logic
• environment_agent/: Environment Agent implementation
• environment_agent/planning_backend.py: default mock backend and backend boundary
• api/index.py: FastAPI serverless entrypoint
• API.md: callable API contract

Local Development Only

Use local ports only when developing or debugging on your machine:

pip install -r requirements.txt
python -m uvicorn api.index:app --reload --port 8765

Then open http://127.0.0.1:8765/api/docs.