POST · Chat · mnueron docs

POST /api/chat

Memory-augmented chat completion. Mnueron pulls the top-8 BM25-relevant memories plus up to 4 matching runbooks into the system prompt, then forwards the conversation to the configured LLM provider (Anthropic by default).

⚠️ Paid plan required, unless the call carries a byok_anthropic_key or byok_openai_key.

curl -X POST https://www.mnueron.com/api/chat \
  -H "Authorization: Bearer $MNUERON_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      { "role": "user", "content": "Help me deploy to prod" }
    ],
    "namespace": "work"
  }'

Body

Field	Type	Required	Notes
`messages`	array	yes	`{ role: "user"
`session_id`	uuid	no	If set, turns are appended to the session
`namespace`	string	no	Constrain memory retrieval to one project
`byok_anthropic_key` / `byok_openai_key`	string	no	Caller's own LLM key

Response

{
  "reply": "Here's how I'd ship this — based on your existing 'Deploy to prod' runbook…",
  "citations": [
    { "id": "01HQ…", "type": "memory", "title": "Prefer feature flags for risky deploys" },
    { "id": "01HR…", "type": "procedural", "title": "Deploy to production" }
  ],
  "runbook_proposal": null,
  "session_id": "01HQ…",
  "via": "anthropic"
}

citations lists every memory or runbook the model consumed. Use these ids in /api/recall to feed the recall-quality loop.
runbook_proposal is non-null when the model offered to save the conversation as a new runbook.
via is the underlying provider (anthropic, openai, or byok).

Sessions

GET /api/chat/sessions / POST /api/chat/sessions / GET|PATCH|DELETE /api/chat/sessions/:id — list, create, fetch, rename, or delete persistent chat sessions. See /docs/api-reference/chat-sessions.