One library. Every workflow.
Anything that can send an HTTP request can read from and write to your library. Your memory becomes the single source of truth the rest of your stack references.
Get an API key
Generate a personal key under Settings → API. Send it on every request as an x-mythos-key header. Keys are scoped to your account and individually revocable, with usage tracking.
curl -H "x-mythos-key: mtk_•••••••" \ "https://mythos.one/api/internal/memos?tag=memory"
The agent API
List, search, read, create, and update memos over JSON. The full machine-readable contract lives at /api/openapi.
GET /api/internal/memos— list and filter your memos (by tag, search, date).POST /api/internal/memos— create a memo from markdown.PATCH /api/internal/memos— update an existing memo.
Requests are rate-limited to 60 per minute per key.
Zapier, n8n, Make
Use the built-in HTTP/webhook step in your automation tool with the endpoints above: pull memos as context before a step runs, or write a memo when a flow finishes. No dedicated plugin required — it's just JSON over HTTPS.
Email-to-memo ingestion
On Scholar and Oracle, your library has an ingestion address — forward an email and it lands as a memo. Configure it under Settings → Ingestion.
MCP and self-discovery
AI clients like Claude, ChatGPT, and Cursor connect natively over MCP — see the MCP guide. Autonomous agents can discover everything from /llms.txt, which indexes the API, the MCP server, and your public content.