Advance a typed builder conversation without generating or writing.
Collects three bounded HumanQuery-compatible clarification rounds and returns the existing synthesis request shape. State is client-carried and strictly revalidated on every turn; this endpoint never invokes an LLM, dispatches work, or persists a proposal.
AuthorizationBearer <rsk_live_*>Rensei API key presented as Authorization: Bearer rsk_live_. Required API-key scopes are recorded per operation in x-rensei-required-scopes.
projectId*stringapplication/json- body
action*stringgoal*string3 <= length <= 4000target?string"workflow""workflow""agent_card"workflowId?string1 <= length <= 120agentCardId?string^([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[1-8][0-9a-fA-F]{3}-[89abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}|00000000-0000-0000-0000-000000000000|ffffffff-ffff-ffff-ffff-ffffffffffff)$uuidNext typed query or completed synthesis request.
application/json- response
status*string"question""ready"state*query?unknownsynthesisRequest?unknownconversation?unknowncurl -X POST 'https://app.rensei.ai/api/projects/string/authoring/conversation' \ -H 'Content-Type: application/json' \ -d '{ "action": "start", "goal": "string"}'{ "status": "question", "state": { "kind": "builder-conversation", "schemaVersion": "1", "goal": "string", "target": "workflow", "workflowId": "string", "agentCardId": "2a5965de-58ff-478a-b9b7-851d0b90b5c2", "nextRound": "outcome", "turns": [ { "round": "outcome", "prompt": "string", "answer": { "property1": null, "property2": null }, "answeredAt": "2019-08-24T14:15:22Z" } ] }, "query": null, "synthesisRequest": null, "conversation": null}Store a reviewed synthesis candidate as an isolated proposal.
Revalidates typed, base, candidate, vocabulary, policy, and proposal-version digests before appending an immutable proposal version. This route never mutates runtime workflows or agent cards.
Simulate and explain a proposed new canvas against one exact Cedar policy revision.
Read-only, bounded author-time simulation for a proposed new canvas. It resolves node/tool resources from the tenant and project vocabulary, loads one exact active workspace/project policy snapshot, evaluates without runtime audit writes, and denies closed on no-match, unknown-node, or evaluation errors. Existing-workflow edit authorization is intentionally outside this create-only surface.