Kanaeru Developer Documentation
Kanaeru exposes a programmatic surface for developers and AI agents alongside the Kana web client. This page is the index: the machine-readable contract lives in the OpenAPI specification at /openapi.json, and a crawler-friendly overview of the whole site lives at /llms.txt. Note the access model: Kanaeru is in an invite-based closed beta with no self-serve signup and no published pricing — every credentialed surface below is provisioned from inside a signed-in, invited workspace. There is no standalone public API program yet.
REST API
The orchestrator API is served from this origin under /orc/. GET /orc/health reports service status as JSON with no authentication. POST /orc/api/automation-triggers/{slug} triggers a workspace automation using an x-api-key header; keys are issued per automation in the Kana automations UI, so triggering one requires an invited workspace that has configured it. All API errors are structured JSON with statusCode, error, and message fields — never HTML.
The full request and response schemas, parameter types, and error shapes are published in OpenAPI 3.1 format at /openapi.json.
MCP server
Agents can connect to the Kanaeru orchestrator over the Model Context Protocol on behalf of an invited account. The MCP endpoint is served at /orc/mcp on this origin, secured by OAuth 2.1 with PKCE; authorization server metadata is discoverable at /orc/.well-known/oauth-authorization-server. Connect an MCP-capable client (Claude, Codex, or any MCP client) to browse visions, outcomes, and tasks, and to drive outcome delivery programmatically.
Authentication
The web client authenticates with Google or Microsoft via Supabase Auth; access is by invitation, and there is no self-serve signup. CLI and MCP access are set up through that same sign-in — the Kana UI walks you through the OAuth consent flow. Automation endpoints authenticate with per-automation API keys sent as the x-api-key header, issued from the signed-in workspace. Web API routes under /api return JSON 401 errors when a session is missing.
Agent resources
Machine-readable entry points: /llms.txt (site overview and when-to-use guidance), /openapi.json (API contract), /sitemap.xml (indexable pages), /robots.txt (crawl policy). The public pages on this site (home, sign-in, about, contact, privacy, docs) also serve a Markdown representation via content negotiation: request it with an "Accept: text/markdown" header.