Skip to content
Jinn

Engines and Limits API

Use these authenticated read routes before sending session overrides.

Method and path Response
GET /api/engines {default,engines}. Each engine registry entry includes availability, default model, effort mechanism, and model capability records.
GET /api/engine-limits?engine=name {generatedAt,default,engines}. The optional filter requests one engine snapshot.
Terminal window
curl -sS "$JINN_GATEWAY_URL/api/engines" \
  -H "Authorization: Bearer $JINN_GATEWAY_TOKEN"
curl -sS "$JINN_GATEWAY_URL/api/engine-limits?engine=claude" \
  -H "Authorization: Bearer $JINN_GATEWAY_TOKEN"

GET /api/engines returns {default,engines}. Registry available means the configured engine executable was discovered; it does not prove provider authentication. Model IDs and effort levels are exact strings. Omit overrides to use defaults, or select only a combination declared by the registry.

GET /api/engine-limits returns {generatedAt,default,engines}. Limits can be live, cached, static, unsupported, or error snapshots depending on the provider CLI. They are observations, not execution caps; configured session and employee budget fields are not runtime-enforced limits at this source pin.

The mutation/refresh routes are intentionally outside this curated discovery contract. Use the two read routes above whenever configuration may have changed.