🧠 RepoMemory

Project Intelligence Platform for AI Coding Agents

SonarQube for AI Coding Agents — Scan → Audit → Generate → Monitor → Detect Drift → Repair → Learn

Web App ⋅ CLI ⋅ MCP Server ⋅ VS Code Extension ⋅ GitHub Action

---

🎯 The Problem

AI coding agents are only as good as the context they receive. Yet the files that control that context — CLAUDE.md, .claudeignore, AGENTS.md, CURSOR.md — are maintained manually, inconsistently, and with no tooling support. When your code evolves but your context files don't, your AI agent works with outdated information.

RepoMemory solves this. It's the first tool that treats AI context quality as a product — automatically scanning, auditing, generating, and monitoring the intelligence layer of your repositories.

---

🎥 Demo

Scan → Audit → Generate in 5 seconds

---

✨ Features

🔍 Multi-Agent Audit

Score your AI context files across 7 quality dimensions (Architecture, Commands, Conventions, Off-limits, Testing, Deployment, Freshness) for 6 AI coding agents:

Agent	File	Score
	CLAUDE.md	0–100
	.cursor/rules.mdc	0–100
	.windsurf/rules.md	0–100
	GEMINI.md	0–100
	AGENTS.md	0–100
	AIDER.md	0–100

🧠 Core Capabilities

📝 Context Generation Generate complete AI context packs for all major coding agents — CLAUDE.md, AGENTS.md, CURSOR.md, GEMINI.md, AIDER.md, .claudeignore, slash commands, and git hooks.	🎯 Quality Suggestions Not just scores — actionable fix steps per dimension. "Add a `## Commands` section listing: `npm run dev`, `npm run build`, `npm test`."
🔄 Smart Regeneration Detects which sections of your context files changed (new dependencies, new directories, config changes) and regenerates only what's needed — not the entire file.	⚡ One-Click Fix Single button to generate + apply all context files. From zero to fully documented in one click.
👁️ Advanced Drift Detection Git-aware analysis with risk scoring (low/medium/high/critical), per-file risk assessment, and repair suggestions. Knows when your code outgrew your context.	🏗️ Architecture Discovery Engine Automatically discovers modules, services, data flows, and tech stack. Generates architecture reports, dependency maps, and service relationship diagrams.
📊 Score Timeline & Comparison Track context quality over time with interactive charts. Compare any two scans side-by-side to see which dimensions improved or regressed.	🏷️ Repository Badge Shields.io-style AI Readiness badge for your README. Show the world your project is AI-ready.
📋 Change Summaries AI-generated summaries of recent git commits for agent context. Keeps AI agents informed about what changed without reading full diffs.	📝 ADR Generator Automatically creates Architecture Decision Records from git history, dependency changes, and config changes. Preserves engineering decisions in `adr/` directory.
🧠 Project Brain Persistent knowledge store for your repository — decisions, lessons learned, technical debt, and known bugs. Stored in `.repomemory/brain/` and accessible to AI agents.	🤖 MCP Server (8 tools) Expose RepoMemory as MCP tools: scan, generate, drift, apply, explain, suggest, health, explain_codebase. Any MCP-compatible AI agent can call them mid-session.

🔌 Integrations

Interface	What it does	Quick Start
🌐 Web App	Full-featured dashboard with scan, audit, generate, compare	npm run dev → localhost:3000 · Live demo
💻 CLI	Run scans, generate, and check from terminal	npm run cli scan .
🤖 MCP Server	Call RepoMemory tools from any AI agent	8 tools, stdio transport
🖥️ VS Code	Sidebar panel with project tree + webview audit	Open vscode/, F5
🔄 GitHub Action	Auto-check PRs for context drift	uses: ./github/actions/repomemory-check

---

🚀 Quick Start

# Clone
git clone https://github.com/PraveenJayaprakash-JP/repomemory.git
cd repomemory

# Install & configure
npm install
cp .env.example .env.local   # Add your AI API key

# Start the web app
npm run dev                   # → http://localhost:3000

# Live demo: https://repomemory-omega.vercel.app

# Or use the CLI
npm run cli scan /path/to/your/repo
npm run cli generate /path/to/your/repo --apply

🐳 One-Click Fix

# Generate + apply all context files in one command
npm run cli generate /path/to/repo --apply

🤖 MCP for AI Agents

Add to your opencode.json or claude_desktop_config.json:

{
  "mcpServers": {
    "repomemory": {
      "command": "npx",
      "args": ["tsx", "mcp/src/index.ts"],
      "cwd": "/path/to/repomemory"
    }
  }
}

Available MCP tools:

Tool	Description
scan_repo	Scan + audit a repo folder
generate_pack	Generate context files via AI
check_drift	Detect stale context
apply_pack	Generate + write to repo
explain_repo	Explain repo architecture
suggest_context_updates	Get recommendations
repo_health	Overall health score
explain_codebase	Comprehensive codebase explanation

---

📸 Screenshots

Dashboard — scan history & score timeline	Scan Page — repo scanner & folder picker
Audit Results — 7-dimension score, quality suggestions, architecture graph

---

⚙️ Configuration

Variable	Default	Description
AI_PROVIDER_API_KEY	—	Your AI provider API key
AI_PROVIDER_BASE_URL	https://opencode.ai/zen/go/v1	OpenAI-compatible endpoint
AI_MODEL	deepseek-v4-flash	Model for generation tasks

---

🏗️ Architecture

repomemory/
├── app/              # Next.js 15 — 16 routes across 7 pages
├── lib/              # 15 core modules
│   ├── scanner.ts    # File system traversal + detection
│   ├── auditor.ts    # 7-dimension quality scoring
│   ├── generator.ts  # AI context pack generation
│   ├── drift.ts      # Drift detection + risk scoring
│   ├── graph.ts      # Architecture graph builder
│   ├── discovery.ts  # Deep architecture discovery
│   ├── adr.ts        # ADR generation from git history
│   ├── brain.ts      # Persistent knowledge store
│   ├── changelog.ts  # Git change summaries
│   ├── ai.ts         # OpenAI-compatible API wrapper
│   ├── prompts.ts    # AI prompt templates
│   ├── storage.ts    # File-based persistence
│   ├── export.ts     # ZIP export
│   └── types.ts      # Shared TypeScript types
├── cli/              # CLI tool (3 commands)
├── mcp/              # MCP server (8 tools)
├── vscode/           # VS Code extension
├── .github/          # GitHub Action + CI workflow
└── tests/            # 24 unit tests

---

🔄 CI Integration

Add to your GitHub Actions workflow:

- name: Check AI Context
  uses: PraveenJayaprakash-JP/repomemory/.github/actions/repomemory-check@main
  with:
    ai-api-key: ${{ secrets.AI_PROVIDER_API_KEY }}
    fail-on-drift: true
    min-score: 60

This checks every PR for:

• ✅ AI Readiness score
• ✅ Context drift detection
• ✅ Risk level assessment
• ✅ Automatic PR comments with findings
• ✅ CI/CD enforcement (optional min-score gate)

---

🛠️ Development

npm test              # 24 unit tests
npm run build         # Production build (2.5s, 16 routes)
npm run cli -- scan . # Scan this repo
npm run cli -- generate . --apply  # Generate context for this repo

---

🗺️ Roadmap

• Multi-agent context generation (6 AI agents)
• Advanced drift detection with risk scoring
• Architecture discovery engine
• ADR generator
• Project Brain (persistent knowledge)
• explain_codebase MCP tool
• Public hosted version
• Team dashboard with multi-repo support
• VSCode marketplace release
• GitHub marketplace action

---

🤝 Contributing

PRs welcome! RepoMemory is built with OpenCode Go — the same AI coding agent it helps configure.

1. Fork the repo
2. Create your feature branch: git checkout -b feat/amazing
3. Commit: git commit -m 'feat: add amazing feature'
4. Push: git push origin feat/amazing
5. Open a PR

---

📄 License

MIT © PraveenJayaprakash-JP

---

⭐ Star us on GitHub — it helps more developers discover RepoMemory

GitHub ⋅ Issues ⋅ Discussions