code-auditor-mcp
Enables AI assistants to search, analyze, and understand multi-language codebases by providing indexed code intelligence via MCP.
README
Code Auditor: AI-Powered Multi-Language Code Intelligence
Your AI understands your code across languages. Code Auditor indexes your entire codebase (TypeScript, JavaScript, and Go) and provides real-time analysis that AI assistants like Claude can actually use to help you write better code.
The Problem It Solves
AI coding assistants are powerful, but they're flying blind. They can't search your codebase, don't know your patterns, and miss critical context. Code Auditor changes that by creating a searchable index of every function, component, and pattern in your code.
How It Works
- Index - Automatically catalogs functions, React components, Go structs, and dependencies
- Analyze - Detects SOLID violations, code duplication, and security issues across multiple languages
- Connect - AI assistants access your code index via MCP (Model Context Protocol)
- Iterate - Get intelligent suggestions based on your actual codebase patterns
Quick Start (2 minutes)
# Add to your project with Claude Code CLI
claude mcp add code-auditor -- npx code-auditor-mcp
# That's it! Now ask Claude:
# "What authentication functions exist in my codebase?"
# "Find all API endpoints and check for rate limiting"
# "Show me Go structs that handle user data"
# "Compare TypeScript and Go implementations of the same feature"
Core Features That Matter
π Multi-Language Code Search
"Find all functions that validate user input"
"Show me where we're calling the payment API"
"What Go structs implement the User interface?"
"Compare error handling patterns between TypeScript and Go"
π― Smart Code Analysis
- SOLID Principles - Catch architecture issues before they spread
- DRY Violations - Find duplicate code that should be refactored
- Security Patterns - Verify auth, rate limiting, SQL injection protection
- Dead Code - Identify unused imports and functions
π€ AI Tool Integration
Auto-generates configurations for:
- Claude (via MCP)
- Cursor
- Continue
- GitHub Copilot
- 10+ other AI assistants
βοΈ Persistent Configuration
Set your analyzer preferences once:
You: "Set SOLID analyzer to allow 3 responsibilities for components"
Claude: Configuration saved! All future audits will use this setting.
π Project task queue (MCP)
Use the project_tasks tool to keep a per-project task list in the local database (titles, status, priorities, due dates, blockers, related files/symbols, and more). Tasks survive sync_index reset: clearing the analysis index removes indexed functions, cached audits, code maps, and schema overlays so you do not keep βghostβ references to deleted codeβit does not delete your task list or analyzer configs.
Real Examples
Example 1: Finding Authentication Patterns
You: "Show me all authentication-related functions"
Claude: Found 23 functions across 8 files:
- `validateToken()` in auth/tokens.ts:45
- `requireAuth()` in middleware/auth.ts:12
- `checkPermissions()` in auth/permissions.ts:78
...
Example 2: Analyzing Code Quality
You: "Audit the user service for issues"
Claude: Found 3 critical issues:
- Single Responsibility violation: UserService handles both auth and profile updates
- SQL injection risk: Raw query in getUserByEmail() at line 234
- Missing rate limiting on password reset endpoint
Example 3: Discovering Patterns
You: "Find React components similar to DataTable"
Claude: Found 4 similar components:
- `UserTable` - extends DataTable with user-specific columns
- `OrderGrid` - implements similar pagination pattern
- `ProductList` - uses same filtering approach
Installation Options
Published package (npm / pnpm / yarn)
npm install -g code-auditor-mcp
# or: pnpm add -g code-auditor-mcp
code-audit # Run analysis
Project install
npm install --save-dev code-auditor-mcp
npx code-audit
Developing from this repository
Requires Node.js 18+. From the app/ directory:
pnpm install
pnpm run build
pnpm test
The package manager in use is pnpm (see packageManager in package.json). Use pnpm run test:parity only if you need the legacy vs universal analyzer parity suite.
CI/CD Integration
# GitHub Actions
- name: Code Audit
run: npx code-audit --fail-on-critical
MCP server: where the index is stored
The Loki index file defaults to <current working directory>/.code-index/index.db. If your MCP host runs with an unexpected or shared cwd, point at a dedicated folder:
- Environment variable
CODE_AUDITOR_DATA_DIRβ directory to use as the storage root (absolute or relative to the process cwd when the server starts). The database file is<resolved>/index.db(no extra.code-indexsegment). - CLI β
node dist/mcp-index.js --data-dir /path/to/data(equivalent to setting the env var). The published binarycode-auditor-mcpalso accepts--data-dirbecause it loads the same bootstrap.
Example Cursor .cursor/mcp.json fragment:
{
"mcpServers": {
"code-auditor": {
"command": "node",
"args": ["/absolute/path/to/code-auditor/app/dist/mcp-index.js", "--stdio"],
"env": {
"CODE_AUDITOR_DATA_DIR": "/Users/you/Library/Application Support/code-auditor"
}
}
}
}
Key Commands
code-audit # Full analysis with HTML report
code-audit -f json # JSON output for CI/CD
code-audit -a solid,dry # Run specific analyzers
code-audit --health # Quick health score (0-100)
The Feedback Loop
- Write Code β Code Auditor indexes it automatically
- Ask AI β "Is there a function to validate emails?"
- Get Context β AI finds
validateEmail()and similar patterns - Improve β AI suggests using existing validation instead of duplicating
- Repeat β Your AI gets smarter about YOUR codebase
Configuration
Set custom thresholds for your architecture:
// Via MCP
mcp.set_analyzer_config({
analyzerName: "solid",
config: {
maxUnrelatedResponsibilities: 4,
patternThresholds: {
"Dashboard": { maxResponsibilities: 6 }
}
}
});
Advanced Search Operators
| What You Want | Search Query |
|---|---|
| Complex functions | complexity:>10 |
| Go functions only | lang:go |
| TypeScript components | lang:typescript component:functional |
| Undocumented exports | exported:true jsdoc:false |
| React hooks usage | component:functional hook:useState |
| Go struct methods | lang:go entity:struct |
| Find dependencies | calls:validateUser |
| Cross-language patterns | name:validateEmail |
| Unused imports | unused-imports file:src |
Performance
- Indexes 10,000+ functions in seconds
- Incremental updates on file changes
- LokiJS for fast in-memory search
- FlexSearch for intelligent queries
Contributing
See CONTRIBUTING.md for guidelines.
License
MIT - Use it anywhere, anytime.
Ready to give your AI x-ray vision into your code?
claude mcp add code-auditor -- npx code-auditor-mcp
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.