Log inSubmit server

Phaedo MCP README

The full upstream README, mirrored for reference. Install config, tool schemas, adoption signals and our own overview live on the Phaedo MCP listing page.

Back to Phaedo MCP View source on GitHub

Phaedo — your cognitive fingerprint, on every AI surface

npm license

Phaedo learns how you think — communication style, working style, decision pattern — into a portable, user-owned cognitive fingerprint, and carries it to any AI tool. This repo is the open half of that system:

  • spec/ — Anamnesis (formerly "Phaedo Protocol"), the open cognitive-fingerprint standard: machine-readable JSON Schemas (fingerprint layers, consultation request/response, deference policies, receipts) + the conformance suite.
  • docs/protocol/ — the Anamnesis spec (v0.1), the MCP binding, and accepted proposals.
  • mcp/ — the reference MCP server (phaedo-mcp on npm): injects your fingerprint at session start and answers agent consultations — an agent at a decision point can ask "would my human proceed here?" and get a calibrated proceed / clarify / escalate / decline signal (or an honest abstain), gated by user-authored deference policies, with an encrypted on-device audit trail. On a blocking signal it can escalate to your paired phone in real time for a live approve / deny / modify, and your overrides teach the act-as-me channel — how you want delegated work handled, not just how you work yourself.

Quick start

Claude Code — install the plugin (it brings the server and the hooks):

Terminal
claude plugin marketplace add phaedo-labs/phaedo-mcp
Terminal
claude plugin install phaedo@phaedo

Claude Desktop — download desktop/phaedo.mcpb and double-click it.

Anything else that speaks MCP — npx phaedo-mcp runs the server over stdio.

Your full fingerprint comes from the Phaedo phone app, learned on-device from your actual usage. The app isn't open to the public yet (phaedo.so has a link to hear when it is). Without it, the Claude Code plugin builds a starter profile on your computer from your own sessions. With neither, the server says plainly that there is no profile; it does not guess. To try it with a made-up example profile, set PHAEDO_FINGERPRINT=sample. See plugin/README.md for what the plugin runs, stores and sends, and mcp/README.md for the full guide (pairing, policies, receipts, the agreement metric).

Design commitments

  • User-owned & local. The fingerprint lives on your devices, encrypted; the server holds no plaintext at rest. Nothing readable is sent to any Phaedo service: with a paired phone, messages between your computer and your phone are end-to-end encrypted, and the relay that carries them off your Wi-Fi can't read them.
  • Signals out, never structure. Consultations return a signal + confidence — never fingerprint internals (spec §10.4).
  • Honest abstention. No coverage → insufficient_signal, not a confident guess.
  • User-authored guardrails win. Deference policies (§10.5) override any inferred signal; standing authorizations are explicit and auditable (§10.6).

What's not in this repo

The learning pipeline that builds fingerprints (conversation capture, on-device extraction models) is Phaedo's proprietary side. This repo is everything a client, agent, or independent producer needs to consume and interoperate: the spec, the schemas, and a complete reference server.

License

Apache-2.0. The Phaedo name and logo are trademarks; see phaedo.so.