io.github.hellosg/coderegistry

The Code Registry MCP Server

Enterprise-grade code intelligence for AI assistants

Quick Start • Features • Use Cases • Documentation • Website

---

Enterprise-grade code intelligence for business leaders by The Code Registry. Now with a native, fully-featured MCP server for AI agents. Analyze codebases for due diligence, tech debt, security vulnerabilities, code quality and more - directly from Claude Desktop, Claude API, and other MCP-compatible clients.

Install as an MCP server

Use this URL when configuring any MCP-compatible client:

https://integrator.app.thecoderegistry.com/api/ai/router

Install as an agent skill

Some MCP clients and editors (including Cursor) require MCP server configuration before the skill can use tools.

Cursor MCP server configuration (example)

{
  "mcpServers": {
    "coderegistry": {
      "url": "https://integrator.app.thecoderegistry.com/api/ai/router"
    }
  }
}

Use this JSON in your Cursor MCP settings, or copy the same example from integrations/cursor/config.json.

Cursor / CallMcpTool known limitation and workaround

Some Cursor builds (and other clients that wrap MCP tools behind a generic CallMcpTool) may show an incomplete tool schema that omits arguments, even though the runtime accepts it.

For The Code Registry tools, always pass required parameters in the arguments field when using CallMcpTool.

Prompt hint that often fixes this behavior:

Use the CallMcpTool with the arguments field to pass the required parameters.

Example wrapper call shape:

{
  "server": "coderegistry",
  "toolName": "create_project",
  "arguments": {
    "user_id": "<user_id>",
    "name": "My Project"
  }
}

Then install the skill:

npx skills add The-Code-Registry/mcp-server

The code-registry skill gives your client AI agent extra context and instructions for using this MCP server correctly, including:

• the recommended tool flow (create_account -> create_project -> create-code-vault -> polling)
• completion rules for first analysis vs re-analysis (snapshot vs comparison reports)
• retry/backoff behavior and LOCAL_AGENT follow-up guidance

What is this?

This is a hosted MCP server that lets AI agents create accounts, manage projects, analyze code, and retrieve analysis results without leaving the conversation. It is designed for business leaders (M&A professionals, CTOs, VCs, board members) who need fast, reliable code insights and clean reports.

Any codebase can be analyzed and results can be ready in as quick as 30 minutes depending on code volume. PDF reports containing the results are automatically emailed to the user after analysis is complete, and as their AI agent you can retrieve the results as soon as they are ready, to show in your conversation with the user or to generate graphs/charts/reports with etc.

What is MCP?

The Model Context Protocol (MCP) is an open standard that enables AI assistants to securely connect to data sources and tools. Learn more at modelcontextprotocol.io.

MCP Protocol Support

This server supports the full MCP JSON-RPC 2.0 flow, including:

• initialize
• tools/list
• tools/call
• resources/list and resources/read
• prompts/list and prompts/get

Prerequisites

• An MCP-compatible client (Claude Desktop, Claude Code, or custom implementation)
• For LOCAL_AGENT: Docker installed on your machine
• For GIT sources: Git repository URL with appropriate access
• Basic understanding of code analysis and software metrics

Existing accounts

If the user already has a Code Registry account, they can generate an API key in the web app and use it directly via the X-API-Key header (preferred). If the MCP client cannot set custom headers, pass api_key in tool arguments instead.

Features

• Zero setup: agents can create accounts automatically, all you need is their email address, name and team/company name
• Privacy-first: Using the LOCAL_AGENT code source type keeps code on your machine and only sends an anonymous zip file of results
• Plan-aware analysis: free accounts can sync code and retrieve summary-level results; paid plans unlock full findings data and premium outputs
• Executive-friendly reports: PDF reports automatically emailed to the user, ready for non-technical stakeholders
• Multiple sources: local repos or folders of code, GIT repositories, or file archives

Sample prompts for your AI agent or system

Once the MCP server is configured, you can ask your AI agent or system questions like:

• "Use The Code Registry to analyze this codebase and present the results in a nice way."
• "Re-analyze this project's code with The Code Registry to show me what's changed since the last analysis."
• "Get the latest analysis for Project X from my Code Registry account and present them to me (here's my API key: XXX)."

MCP Protocol Quick Start

0) Initialize (recommended)

{
  "jsonrpc": "2.0",
  "method": "initialize",
  "params": {
    "protocolVersion": "2024-11-05",
    "capabilities": {}
  },
  "id": 1
}

1) Create account

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "create_account",
    "arguments": {
      "email": "user@example.com",
      "name": "User Name",
      "team_name": "Example Org",
      "integrator_id": "agent-name"
    }
  },
  "id": 2
}

Response: Returns api_key, team_id, and user_id (inside result.content[0].text as JSON) - store the API key securely and include it in all subsequent requests via X-API-Key header.

2) Create project

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "create_project",
    "arguments": {
      "user_id": "<user_id>",
      "name": "My Project",
      "description": "Optional"
    }
  },
  "id": 3
}

3) Create code vault (recommended: LOCAL_AGENT)

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "create-code-vault",
    "arguments": {
      "project_id": "<project_id>",
      "user_id": "<user_id>",
      "name": "My Code Vault",
      "source_type": "LOCAL_AGENT"
    }
  },
  "id": 4
}

Response: Includes next_steps.commands with Docker commands to run locally.

4) Poll for results

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "get-code-vault-results",
    "arguments": { "vault_id": "<vault_id>" }
  },
  "id": 5
}

Note: Results are typically ready in 30 minutes to a few hours, depending on codebase size. Poll with exponential backoff.

5) Fetch report URLs

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "get-code-vault-reports",
    "arguments": { "vault_id": "<vault_id>" }
  },
  "id": 6
}

Note: PDF reports are generated after analysis completes and are automatically emailed to the user.

Re-analyze an existing code vault

To re-run analysis on an existing code vault:

1. Call reanalyze-code-vault
2. Run the LOCAL_AGENT again if the original source type was LOCAL_AGENT
3. Poll the same summary/results/report tools as usual

Important: Once re-analysis starts, get-code-vault-summary, get-code-vault-results, and get-code-vault-reports return the new version only. Previous version data is no longer accessible via these tools.

Use Cases

M&A Due Diligence

Quickly assess acquisition targets for technical risk, security vulnerabilities, and architecture concerns. See examples/use-cases/due-diligence.md.

Tech Debt Analysis

Monitor technical debt across your portfolio and track improvements over time. See examples/use-cases/tech-debt-analysis.md.

Security Audit

Identify security vulnerabilities and compliance issues before they become problems. See examples/use-cases/security-audit.md.

Portfolio Monitoring

Track engineering metrics across multiple companies in your investment portfolio. See examples/use-cases/portfolio-monitoring.md.

Configuration

See integrations/ for client-specific setup:

• Claude Desktop - Desktop app configuration
• Claude API - API integration with Python/TypeScript examples
• Cline - VS Code extension setup
• Cursor - Cursor IDE integration
• Custom agents - Build your own MCP client

Agent Skill (Optional)

This repository includes a reusable agent skill at skills/code-registry that gives agents additional context and concrete operating instructions for this MCP server (tool order, polling lifecycle, completion rules, and result interpretation).

• Skill entry point: skills/code-registry/SKILL.md
• Lifecycle details: skills/code-registry/references/vault-lifecycle.md
• Interpretation guidance: skills/code-registry/references/result-interpretation.md
• Polling helper script: skills/code-registry/scripts/poll_vault_status.py

Install the code-registry folder into your agent's skills directory, then run your agent normally:

• Codex: $CODEX_HOME/skills/
• Claude Code: ~/.claude/skills/
• Cursor project-local: .cursor/skills/

See skills/README.md for usage details and script examples.

Pricing

• Free Tier: Up to 100,000 lines of code. Includes account creation, project/vault management, code sync, and summary-level MCP outputs.
• Paid Plans: Unlock full analysis data and premium features such as verification badges, public verification pages, Code Score (out of 1,000), and other advanced capabilities. Paid plans also enable the full-data MCP tools: get-security-issues, get-code-smells, get-git-history, get-contributors, get-components, get-code-iq-automated-queries, and get-the-code-score. Verification-only (VerifyYourCode) plans do not include access to these tools.
• Enterprise: Contact us at thecoderegistry.com for custom pricing and volume discounts.

Available on all plans:

• Account and API key management
• Project and code vault lifecycle tools
• LOCAL_AGENT, GIT, and FILE_ARCHIVE ingestion

Documentation

• Getting started: docs/getting-started.md - Step-by-step walkthrough
• API reference: docs/api-reference.md - Complete action reference
• Authentication: docs/authentication.md - API key management
• Facets glossary: docs/facets.md - Definitions for analysis facets in results
• Troubleshooting: docs/troubleshooting.md - Common issues and solutions
• Architecture: docs/architecture.md - System design overview
• Skills guide: skills/README.md - Install and use the Code Registry agent skill

Why Use The Code Registry MCP Server?

For M&A Professionals:

• Fast technical due diligence (hours, not weeks)
• No manual code review needed
• Executive-friendly PDF reports
• Objective technical risk assessment

For CTOs & Technical Leaders:

• Monitor technical debt across portfolio
• Track security vulnerabilities
• Understand codebase complexity
• Make data-driven technical decisions

For VCs & Board Members:

• Assess technical risk in portfolio companies
• Track engineering metrics over time
• Get objective technical insights
• Support portfolio company CTOs

Support

• Email: support@thecoderegistry.com
• Issues: Use GitHub Issues for bug reports and feature requests
• Website: thecoderegistry.com

License

See LICENSE file for details.

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

---

Built with ❤️ by The Code Registry

The World's first code intelligence platform built for business leaders