Reference
API Reference
The BRICQS REST API. All endpoints return JSON. Authentication is via Bearer token obtained from POST /auth/login.
Base URL
https://api.bricqsai.comAuthentication
All endpoints (except /health and /status) require an Authorization header:
Authorization: Bearer <access_token>POST /auth/login
bashcurl -X POST https://api.bricqsai.com/auth/login \
-H "Content-Type: application/json" \
-d '{"email": "you@example.com", "password": "your-password"}'json{
"access_token": "eyJ...",
"refresh_token": "...",
"token_type": "bearer",
"expires_in": 900
}Deployments
GET
/deployments/listList your deployments. Optional ?environment= filter.
POST
/deployments/deployCreate a new deployment.
GET
/deployments/{id}Get a single deployment by ID.
POST
/deployments/{id}/stopStop a running deployment (scale to zero).
DELETE
/deployments/{id}Permanently delete a deployment.
GET
/deployments/{id}/logsGet container logs. ?lines=100
GET
/deployments/{id}/metricsReal-time BRICQS metrics. ?hours=6
POST
/deployments/{id}/promotePromote preview → production with zero downtime.
GET
/deployments/modelsList the deployable model catalog.
GET
/deployments/summaryAggregated stats: active count, GPU hours, failure rate.
Deploy request body
namestringrequiredName for this deployment. Used to generate the BRICQS deployment name.
model_idstringrequiredA model ID from GET /deployments/models.
environment'production' | 'preview' | 'development'Deployment environment.
min_replicasintegerMinimum replicas. 0 = scale to zero when idle.
max_replicasintegerMaximum replicas for auto-scaling.
Models / Runtimes
POST
/runtime/{project_id}/chat/completionsOpenAI-compatible LLM proxy for a project.
POST
/agent/runRun a tool-using agent with up to 6 reasoning steps.
POST
/rag-project/queryQuery a RAG-indexed document collection.
POST
/workflow/{project_id}/runRun a multi-step AI workflow.
Platform
GET
/statusPublic platform status. No auth required.
GET
/healthAPI health check. No auth required.
POST
/support/ticketsSubmit a support ticket.
Errors
All errors return { "detail": "message" } with the appropriate HTTP status:
400Bad request — invalid parameters401Missing or invalid access token403Insufficient permissions (account disabled, not admin, etc.)404Resource not found409Conflict — e.g. duplicate deployment for environment500Internal server error — check logs