Gateway HTTP reference

The gateway exposes a control plane and runs surface over HTTP. This page is the human-readable index; the OpenAPI spec is the machine-readable source of truth.

Auth

All routes require a bearer token unless explicitly noted. Configure tokens via gateway config or environment. Unauthenticated requests get 401.

Health & metrics

Method	Path	Description
GET	/health	Liveness probe; returns {status}
GET	/metrics	Prometheus-format metrics

Gateway lifecycle

Method	Path	Description
POST	/gateway/pause	Stop accepting new runs
POST	/gateway/resume	Resume accepting new runs
POST	/gateway/reload	Re-read pipelines and config

Pipelines

Method	Path	Description
GET	/pipelines	List registered pipelines
GET	/pipelines/:id	Describe one pipeline
POST	/pipelines/:id/run	Run a pipeline synchronously
POST	/pipelines/:id/start	Start an async run; returns runId

Runs

Method	Path	Description
GET	/runs/:runId	Get run record
GET	/runs/:runId/events	SSE/JSONL event stream for a run
POST	/runs/:runId/resume	Resume a waiting run
POST	/runs/:runId/approve	Approve a paused approval gate
POST	/runs/:runId/deny	Deny a paused approval gate

Approvals

Method	Path	Description
GET	/approvals	List pending approvals

Triggers

Method	Path	Description
GET	/triggers	List registered triggers
POST	/triggers/:id/fire	Manually fire a trigger

Schedules

Method	Path	Description
GET	/schedules	List schedules
POST	/schedules	Add a schedule
DELETE	/schedules/:id	Remove a schedule

Sessions

Method	Path	Description
GET	/sessions	List ACP sessions
GET	/sessions/:id	Get one session
POST	/sessions/:id/resume	Resume a session
POST	/sessions/prune	Prune expired/old sessions
DELETE	/sessions/:id	Drop a session

Debug

Method	Path	Description
GET	/debug/breakpoints	List active breakpoints
POST	/debug/breakpoints	Add a breakpoint
DELETE	/debug/breakpoints/:stepId	Remove a breakpoint
GET	/debug/runs	List runs paused at breakpoints
POST	/debug/runs/:runId/release	Release a paused run

OpenAI compatibility (optional surface)

Method	Path	Description
POST	/v1/chat/completions	OpenAI-compatible chat surface
POST	/v1/responses	OpenAI-compatible responses API

See also

• docs/reference/openapi.yaml — full request/response schemas
• docs/reference/cli.md — the skelm CLI uses these endpoints