Local MCP server that injects a user's Phaedo fingerprint projection into MCP clients, and also provides consultation and authorization-check tools based on that fingerprint. It can source the fingerprint from a paired phone or from local JSON, and only operates over local stdio in standard mode.
From the registry: Carry your Phaedo cognitive fingerprint into any MCP client: style injection + decision consults.
$ npm run pair$ ollama pull llama3.1:8bPlease install the `phaedo-mcp` MCP server into my current AI client (that's you).
Required prerequisites (do these first if not already done):
- **Install Node.js** — Node.js is required to run the MCP server. (https://nodejs.org/)
Optional prerequisites:
- Optional: pair the server with your phone — For the preferred live fingerprint source, create a pairing record by running the pairing flow. Without pairing, the server falls back to local JSON sources and can still run using the bundled sample fingerprint. Run: `npm run pair`
- Optional: install or run a local Ollama model for consult features — The model judge for consultation is enabled by the installer when a local Ollama/OpenAI-compatible endpoint is available. If you want consult model support, run a general chat model locally. Run: `ollama pull llama3.1:8b`
Canonical MCP server config (stdio transport):
- command: `npx`
- args: ["phaedo-mcp"]
- optional environment variables:
- `PHAEDO_FINGERPRINT`: Path to a local fingerprint JSON vault payload. Used for Phase 1a fallback/demo when no pairing record is configured. (example: `<path-to-fingerprint.json>`)
- `PHAEDO_PAIRING`: Path to the pairing record JSON used to pull the live fingerprint from the paired phone. If omitted, the server also looks for pairing.json in the project defaults described by the README. (example: `<path-to-pairing.json>`)
- `PHAEDO_MCP_STATE_DIR`: Directory for server state such as the encrypted at-rest cache, key file, and receipts. (example: `<path-to-state-dir>`)
- `PHAEDO_RECEIPTS_CAP`: Maximum number of encrypted consultation receipts to retain. (example: `<max-receipt-count>`)
- `PHAEDO_CONSULT_MODEL`: Model name for the local/OpenAI-compatible consult model judge. (example: `<model-name>`)
- `PHAEDO_CONSULT_MODEL_URL`: Base URL for the local or OpenAI-compatible endpoint used by the consult model judge. (example: `<openai-compatible-endpoint-url>`)
- `PHAEDO_CONSULT_API_KEY`: API key for OpenAI-compatible consult model endpoints, if required by that endpoint. (example: `<your-api-key>`)
- `PHAEDO_CONSULT_POLICIES`: Path to a consult policies JSON file that overrides the default consult policies location. (example: `<path-to-consult-policies.json>`)
Note: Preferred portable install path is `npx phaedo-mcp` over stdio. The server can run with no env vars using the bundled sample fingerprint, but real usage typically uses either a phone pairing record or a local fingerprint file. README also documents manual local-node execution from a clone and GUI installer scripts, but the canonical MCP shape is stdio command/args/env only.
Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.PHAEDO_MCP_STATE_DIROverrides the local state directory used for key material, cache, and receipts.PHAEDO_PAIRINGPath to the pairing record used to pull the live fingerprint from the paired phone.PHAEDO_FINGERPRINTPath or payload source for the local fingerprint JSON fallback.PHAEDO_RECEIPTS_CAPSets the maximum number of encrypted consultation receipts kept on device.AI orchestration with hive-mind swarms, neural networks, and 87 MCP tools for enterprise dev.
Persistent codebase knowledge graph. Survives session restarts and context compaction.
A powerful toolkit for coding, providing semantic retrieval and editing capabilities.