LaunchChair Agent API docs.
The Agent API lets an external runner read project context, claim queued work, complete runs, and let LaunchChair apply, validate, retry, or remediate structured output.
Authentication
Private Agent API calls use bearer token auth.
Create an agent token inside Settings -> API Access. Send it on private Agent API requests with the Authorization header. Public documentation routes do not require this token, but project reads, run claims, run completion, exports, and writes always do.
curl -sS \ -H "Authorization: Bearer <full lc_at token from Settings API Access>" \ https://www.launchchair.io/api/agent/v1/projects/PASTE_PROJECT_UUID/spec
Tokens are shown once when created.
Tokens can be scoped to a workspace and optionally one or more projects.
Revoked or expired tokens return 401.
Private agent requests are audited and rate-limited.
Public catalog
Agents can discover the API shape without a login.
Use the public catalog for generic endpoint, scope, bridge, and MCP tool discovery. It intentionally returns placeholders and docs only. It does not return token metadata, workspace data, project specs, queued runs, logs, or user content.
curl -sS https://www.launchchair.io/api/agent/v1/docs
Loop endpoint
The project loop endpoint queues the next actionable step.
The loop endpoint inspects project state, can save the first Idea Snapshot from an ideaSnapshot payload, queues the next agent-actionable bucket or build card, and returns phase synthesis links when completed phases are detected. MCP clients usually call this through launchchair_continue_project instead of calling the HTTP endpoint directly.
curl -sS -X POST \
-H "Authorization: Bearer <full lc_at token from Settings API Access>" \
-H "Content-Type: application/json" \
https://www.launchchair.io/api/agent/v1/projects/PASTE_PROJECT_UUID/loop/continue \
-d '{"stopPhaseKey":"seo_technical","ideaSnapshot":{"projectName":"PASTE_PROJECT_NAME","problem":"PASTE_THE_USER_PAIN","targetUser":"PASTE_THE_TARGET_USER","solution":"PASTE_THE_PROPOSED_SOLUTION"}}'Endpoint catalog
Private Agent API endpoints
These routes require bearer token auth except for the public docs endpoint. The table is safe to publish because it contains route patterns and scopes, not project data.
| Method | Path | Scope | Risk | Description |
|---|---|---|---|---|
| POST | /api/agent/v1/heartbeat | prompts.read | read | Send heartbeat + client metadata for agent presence. |
| POST | /api/agent/v1/runs/next | prompts.read | read | Claim next queued run for the current token. |
| POST | /api/agent/v1/runs/:runId/complete | prompts.read | write | Complete/fail the claimed run, send structured result payload, and let LaunchChair apply, validate, retry, or queue remediation when configured. |
| POST | /api/agent/v1/projects/:projectId/loop/continue | prompts.read | write | Inspect the project, optionally save the first Idea Snapshot from ideaSnapshot, queue the next agent-actionable bucket or build-card run, and return phase synthesis links when completed phases are detected. |
| GET | /api/agent/v1/projects/:projectId/spec | spec.read | read | Read the full normalized project spec. |
| PATCH | /api/agent/v1/projects/:projectId/spec | spec.write | write | Write project spec directly (disabled for MVP token profiles). |
| GET | /api/agent/v1/projects/:projectId/buckets | buckets.read | read | List phase buckets with current status and metadata. |
| GET | /api/agent/v1/projects/:projectId/buckets/:phaseKey/:bucketKey | buckets.read | read | Read a single bucket payload for the target phase/bucket. |
| PATCH | /api/agent/v1/projects/:projectId/buckets/:phaseKey/:bucketKey | buckets.write | write | Write non-prompt bucket data for the target phase/bucket. |
| GET | /api/agent/v1/projects/:projectId/buckets/:phaseKey/:bucketKey/prompt | prompts.read | read | Retrieve compiled prompt text for a bucket. |
| POST | /api/agent/v1/projects/:projectId/buckets/:phaseKey/:bucketKey/apply-spec-update | buckets.write | write | Apply parsed SPEC_UPDATE payload for a bucket run. |
| GET | /api/agent/v1/projects/:projectId/build-cards | build_cards.read | read | List build cards and board state for the project. |
| PATCH | /api/agent/v1/projects/:projectId/build-cards/:cardId | build_cards.write | write | Update build card status/testing metadata. |
| GET | /api/agent/v1/projects/:projectId/build-cards/:cardId/prompt | prompts.read | read | Retrieve compiled prompt text for a build card. |
| POST | /api/agent/v1/projects/:projectId/build-cards/:cardId/apply-run | build_cards.write | write | Apply build card CHANGELOG_JSON/REQUIRES run output. Completion-owned auto-apply is preferred for looped agent runs. |
| GET | /api/agent/v1/projects/:projectId/exports/mvp-ai-spec | exports.read | read | Download AI-ready MVP markdown export. |
| GET | /api/agent/v1/projects/:projectId/exports/phase/:phaseKey | exports.read | read | Download phase summary markdown export. |
Machine-readable summary
What agents should remember
LaunchChair Agent API private routes require Authorization: Bearer <agent-token>.
The public Agent API docs endpoint is /api/agent/v1/docs.
The project loop endpoint is /api/agent/v1/projects/:projectId/loop/continue.
Run completion lets LaunchChair apply, validate, retry, or remediate output according to project state.
FAQ
Does the public API docs endpoint expose private data?
No. It exposes generic endpoint metadata, scope names, bridge file names, MCP tool names, and setup placeholders only.
Should agents call /api/agent/v1 or /api/agent/v1/docs first?
Use /api/agent/v1/docs for public discovery. The authenticated /api/agent/v1 route can return token-specific catalog details after a bearer token is configured.
Agent API docs · MCP setup · structured project loops