name: api-service
description: "Guides SuperAGI FastAPI routes, API-key and JWT auth, webhooks,
migrations, and SQLAlchemy-backed service debugging."
disable-model-invocation: true
metadata:
disco-role: operating
license: MIT
SuperAGI API Service
Use this sub-skill when the task is about SuperAGI's FastAPI service, public API
routes, auth, webhooks, or database-backed service behavior.
Read First
Core Route Groups
SuperAGI's main.py registers many routers under stable prefixes:
/users, /organisations, /projects
/budgets, /agents, /agentexecutions, /agentexecutionfeeds,
/agentexecutionpermissions
/resources, /configs, /toolkits, /tool_configs
/agent_templates, /agent_workflows, /analytics, /models_controller
/google, /twitter, /knowledges, /knowledge_configs
/vector_dbs, /vector_db_indices, /marketplace, /api-keys, /v1/agent
/webhook
Use the API reference for endpoint-level selection; keep the root router map in
mind when a caller names only a resource or action.
Typical Decisions
- Login/API-key path vs GUI session path: Choose the JWT routes in
main.py
for browser/login flows and the API-key protected /v1/agent routes for
external automation.
- API route vs model/tool route: If the task is about a request/response
shape or status code, stay in this sub-skill. If the task is about what a
workflow does after the request, route to
agents-workflows or
toolkits-integrations.
- Schema vs behavior: Use
data-models.md for entity shapes and
inspect_superagi_routes.py for route discovery before making changes or
writing a new reference.
- Migration sensitivity: The app seeds and migration state influence many
endpoints. When the database layout is involved, check whether the user needs
a schema fix or only a controller call.
Safe Workflow
- Use the static route inspector before importing
main.py if the goal is only
to understand endpoint coverage.
- Treat the FastAPI app as side-effectful: importing
main.py can construct a DB
engine and register startup handlers.
- Avoid starting the live API unless the downstream user explicitly wants that
effect and the PostgreSQL/Redis dependencies are available.
Boundary Notes
- Deployment topology, config file values, and container startup belong to
deployment-configuration.
- Tool execution, toolkit registration, and marketplace tool downloads belong to
toolkits-integrations.
- Agent workflow execution, prompts, parsing, and Celery background behavior
belong to
agents-workflows.
- Provider, resource, and vector DB settings that are only needed to satisfy API
requests belong to
models-resources-vector.