Skip to main content
All /v1 endpoints take Authorization: Bearer mk_live_… and are scoped to the key’s workspace. See Introduction for auth, rate limits, and versioning.

​
GET /v1/envs

List the workspace’s environments (mirrors), newest first. 
curl https://api.runmirrors.com/v1/envs \
  -H "Authorization: Bearer mk_live_..."
Returns an array of environment summaries: 
[
  {
    "id": "env_…",
    "name": "airline",
    "slug": "airline",
    "status": "ready",
    "framework": "langgraph",
    "n_traces": 412,
    "n_tools": 9,
    "schema_source": "traces",
    "covered_pct": 88.0,
    "simulated_pct": 12.0,
    "has_agent": true,
    "language": "python"
  }
]

​
GET /v1/envs/{slug}

Fetch one environment by slug. Returns the same summary shape as the list endpoint. Slugs from other workspaces return 404 — never another tenant’s data. 
curl https://api.runmirrors.com/v1/envs/airline \
  -H "Authorization: Bearer mk_live_..."

​
POST /v1/envs/{slug}/query

Run a one-shot query against the mirror and get back the agent’s response plus the full trace. This is the same run path as the Playground: entitlement and sandbox quota gates, deterministic seeding, and per-workspace rate limiting all apply. 
curl -X POST https://api.runmirrors.com/v1/envs/airline/query \
  -H "Authorization: Bearer mk_live_..." \
  -H "Content-Type: application/json" \
  -d '{"input": "Cancel booking BK-4471", "seed": 42}'

​
Body

​
input
string
required
The user message to run against the mirror.
​
seed_instructions
string
Optional world-seeding instructions (e.g. “a customer with one refundable booking”). The mirror’s world is constructed to satisfy them.
​
seed
integer
World seed. The same seed + instructions produce a byte-identical world. Omit to draw a fresh seed — it’s returned in the result so you can replay.
​
auto_seed
boolean
default:"false"
Derive seed instructions from the query itself when none are given.
​
model
string
Model id to run the agent on. Omit for the environment default.

​
Response

{
  "response": "Done — BK-4471 is cancelled.",
  "trace": [
    { "kind": "tool", "label": "cancel_booking", "dur": "0.4s", "detail": "{\"id\": \"BK-4471\"}" },
    { "kind": "ok", "label": "final answer", "dur": "1.1s", "detail": "" }
  ],
  "n_spans": 2,
  "fidelity": { "high": 2, "low": 0 },
  "seed": 42,
  "seeded": true,
  "world_hash": "sha256:…",
  "pinned": { "bookings": ["BK-4471"] },
  "seed_instructions": "a customer with booking BK-4471",
  "seed_note": null
}
The seed, world_hash, and effective seed_instructions are reproducibility receipts — feed the same seed and instructions back in and the world (and run) is identical.

​
GET /v1/usage

Task-minutes used vs. your plan allowance for the workspace — the same numbers as mirrors usage and the dashboard. 
curl https://api.runmirrors.com/v1/usage \
  -H "Authorization: Bearer mk_live_..."