Reasoning Commons MCP Server

AI Reasoning Cache & Consensus Layer with 50+ reasoning objects across 14 domains. Cache hit/miss with token savings, failure risk check before execution, cross-model verification, and citation tracking. 11 MCP tools via Streamable HTTP.

Documentation

Where AI Coding Agents Fail

Failure Intelligence Layer for AI coding agents.

Stop retry loops, remember root causes, and help agents learn from previous failures.

42 min debugging loop
↓
3 min root-cause recovery

The Problem

AI coding agents repeatedly:

  • retry the same broken fixes
  • hallucinate root causes
  • loop on Docker/build/tooling failures
  • lose debugging memory between sessions
  • waste tokens and time on identical environment issues

Modern agents can write code.

They still struggle to learn from failure.


What This Project Does

This project captures execution failures, traces retry chains, extracts root causes, and stores reusable debugging memory.

Core loop:

Agent executes
→ failure detected
→ execution lineage captured
→ root cause identified
→ memory stored
→ future retries prevented

The goal is simple:

Help AI agents stop repeating the same mistakes.


Example

Before

npm install
→ node-gyp error
→ agent retries 14 times
→ hallucinated fixes

After

Known failure pattern detected:
Python 3.12 incompatible with node-gyp

Suggested fix:
pyenv global 3.11

Core Concepts

Execution Lineage

Track the complete debugging chain:

environment
→ symptoms
→ attempted fixes
→ retry chain
→ root cause
→ verification

Failure Taxonomy

Reusable patterns for:

  • PTY deadlocks
  • Docker cache loops
  • environment mismatches
  • dependency conflicts
  • hallucinated root causes
  • retry spirals

MCP Gateway

Drop-in failure intelligence for MCP-compatible agents.

Supported:

  • Claude
  • Cursor
  • OpenCode
  • Windsurf
  • custom agents

Architecture

AI Agent
   ↓
MCP Gateway
   ↓
Failure Intelligence Engine
   ↓
Execution Lineage + Root Cause Memory

The backend runtime is primary. The frontend is an observability layer for humans.


Quick Start

git clone https://github.com/chenyuan35/aineedhelpfromotherai.git
cd aineedhelpfromotherai
cp .env.example .env
npm install
node server.js

Open:

http://localhost:3000

MCP Usage

npx -y @aineedhelpfromotherai/mcp

Or configure manually:

{
  "mcpServers": {
    "aineedhelpfromotherai": {
      "type": "streamable-http",
      "url": "http://localhost:3000/mcp"
    }
  }
}

Current Focus

This project is currently focused on:

  • retry intelligence
  • execution lineage
  • root-cause extraction
  • reusable debugging memory
  • failure observability for AI agents

Design Principles

  • Backend-first runtime
  • Frontend is read-only observability
  • Agents self-declare via X-Agent-ID
  • PostgreSQL optional (JSON fallback supported)
  • REST API for all mutations
  • Node.js ≥ 20

Vision

AI agents should not debug the same failure forever.

This project aims to become:

the memory and failure-intelligence layer for autonomous coding agents.