# nicobailon/pi-mcp-adapter

Token-efficient MCP adapter for Pi coding agent

Repository: https://github.com/nicobailon/pi-mcp-adapter
Canonical: https://ross.abutalabs.com/products/pi-mcp-adapter
Language: TypeScript
License: MIT
License Family: permissive
Topics: ai, claude, coding-agent, extension, llm, mcp, model-context-protocol, pi
Last push: 2026-08-26T17:13:51+00:00

## Health v2 (maintenance only)
Score: 82/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 16
- inputs: {"age_days": 226, "days_push": 7, "days_rel": 7, "gap_med": 1, "n_releases_24m": 58}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1332, forks 292 (observed 2026-08-28T04:04:23.995644+00:00)

## What it is
A token-efficient MCP (Model Context Protocol) adapter extension for the Pi coding agent that exposes a single lightweight proxy tool (~200 tokens) instead of hundreds of verbose tool definitions. It discovers MCP tools on demand, starts servers lazily only when used, and can import existing MCP configs from hosts like Cursor, Claude Code, and Codex.

## Use cases
- use MCP servers with a coding agent without burning context window tokens
- reduce the token cost of verbose MCP tool definitions
- connect MCP tools to the Pi coding agent
- load MCP servers on demand only when tools are actually used
- import existing MCP configs from Cursor, Claude Code, or Codex
- access databases, browsers, and APIs through a single proxy tool

## When to choose
- You use the Pi coding agent and want MCP server access without bloating your context window
- You connect multiple MCP servers and their combined tool definitions consume too many tokens
- You want lazy server startup and on-demand tool discovery instead of always-loaded tools
- You want to reuse existing MCP configuration files across coding agents

## When to avoid
- You don't use the Pi coding agent, since this is a Pi-specific extension
- You need native MCP tool schemas exposed directly to the model rather than via a proxy tool
- You rely on MCP features that require eagerly loaded, fully described tools

## Facets
- artifact type: plugin
- maturity: active
- function: mcp, developer-tools
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: cli, cross-platform
- tags: token-efficiency, context-window, coding-agent-extension, proxy-tool, lazy-loading, on-demand-tool-discovery, config-import, model-context-protocol, ai-agents, command-line, nodejs

## Member repositories
- nicobailon/pi-mcp-adapter (main) score 82

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:23.995644+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:45:13.368184+00:00, confidence not recorded.
  - readme: https://github.com/nicobailon/pi-mcp-adapter (fetched 2026-08-28T04:04:23.995644+00:00, sha e98bd1ca0f63)
  - registry_npm: https://registry.npmjs.org/pi-mcp-adapter (fetched 2026-08-29T12:04:30.033951+00:00, sha 27fe9059e5fa)
- Data as of 2026-08-30T08:39:29.467469+00:00.