Local MCP Server

mycontext mcp starts a Model Context Protocol server over stdio. It runs entirely on your machine — no network, no hosted service, no cost — and exposes mycontext's offline capabilities to any MCP client (Claude Code, Cursor, Cowork, and others).

Install

The MCP server needs the optional mcp extra:

pip install "mycontext-ai[mcp]"

Run

mycontext mcp

The process speaks MCP over stdio and blocks until the client disconnects.

Tools exposed

Tool	LLM call?	Description
suggest_patterns	No	Recommend cognitive patterns for a question, with an optional ordered chain and reasoning
transform	No	Turn raw input into a structured, portable context (assembled prompt + patterns applied)
score_output	No	Heuristically score an LLM output against the context that produced it

All three are deterministic and offline.

suggest_patterns(question, max_patterns=5)

{
  "patterns": [
    {"name": "root_cause_analyzer", "category": "reasoning", "reason": "..."}
  ],
  "chain": ["root_cause_analyzer", "data_analyzer"],
  "reasoning": "..."
}

transform(text, patterns="auto")

{
  "assembled": "ROLE: ...\nTASK: ...",
  "patterns_applied": ["decision_framework"]
}

score_output(context_prompt, output)

{
  "overall": 0.82,
  "dimensions": {"instruction_following": 0.9, "reasoning_depth": 0.8},
  "strengths": ["..."],
  "weaknesses": ["..."]
}

Wiring it into a client

Point your MCP client at the command. For example, in a client that uses a JSON server config:

{
  "mcpServers": {
    "mycontext": {
      "command": "mycontext",
      "args": ["mcp"]
    }
  }
}