DeepSource MCP Server

A Model Context Protocol (MCP) server that integrates with DeepSource to provide AI assistants with access to code quality metrics, issues, and analysis results.

Table of Contents

• Overview
• Quick Start
• Installation
• Configuration
• Available Tools
• Usage Examples
• Architecture
• Development
• Troubleshooting & FAQ
• Contributing
• License
• External Resources

Overview

The DeepSource MCP Server enables AI assistants like Claude to interact with DeepSource's code quality analysis capabilities through the Model Context Protocol. This integration allows AI assistants to:

• Retrieve code metrics and analysis results
• Access and filter issues by analyzer, path, or tags
• Check quality status and set thresholds
• Analyze project quality over time
• Access security compliance reports (OWASP, SANS, MISRA-C)
• Monitor dependency vulnerabilities
• Manage quality gates and thresholds

Quick Start

1. Get Your DeepSource API Key

1. Log in to your DeepSource account

2. Navigate to Settings → API Access

3. Click Generate New Token

4. Copy your API key and keep it secure

2. Install in Claude Desktop

1. Open Claude Desktop

2. Go to Settings → Developer → Edit Config

3. Add this configuration to the `mcpServers` section:

{
  "mcpServers": {
    "deepsource": {
      "command": "npx",
      "args": ["-y", "deepsource-mcp-server@latest"],
      "env": {
        "DEEPSOURCE_API_KEY": "your-deepsource-api-key"
      }
    }
  }
}

4. Restart Claude Desktop

3. Test Your Connection

Ask Claude: "What DeepSource projects do I have access to?"

If configured correctly, Claude will list your available projects.

Installation

NPX (Recommended)

The simplest way to use the DeepSource MCP Server:

{
  "mcpServers": {
    "deepsource": {
      "command": "npx",
      "args": ["-y", "deepsource-mcp-server@latest"],
      "env": {
        "DEEPSOURCE_API_KEY": "your-deepsource-api-key",
        "LOG_FILE": "/tmp/deepsource-mcp.log",
        "LOG_LEVEL": "INFO",
        "RETRY_MAX_ATTEMPTS": "3",
        "RETRY_BASE_DELAY_MS": "1000",
        "RETRY_MAX_DELAY_MS": "30000",
        "RETRY_BUDGET_PER_MINUTE": "10",
        "CIRCUIT_BREAKER_THRESHOLD": "5",
        "CIRCUIT_BREAKER_TIMEOUT_MS": "30000"
      }
    }
  }
}

Docker

For containerized environments:

{
  "mcpServers": {
    "deepsource": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "DEEPSOURCE_API_KEY",
        "-e",
        "LOG_FILE=/tmp/deepsource-mcp.log",
        "-v",
        "/tmp:/tmp",
        "sapientpants/deepsource-mcp-server"
      ],
      "env": {
        "DEEPSOURCE_API_KEY": "your-deepsource-api-key"
      }
    }
  }
}

Local Development

For development or customization:

{
  "mcpServers": {
    "deepsource": {
      "command": "node",
      "args": ["/path/to/deepsource-mcp-server/dist/index.js"],
      "env": {
        "DEEPSOURCE_API_KEY": "your-deepsource-api-key",
        "LOG_FILE": "/tmp/deepsource-mcp.log",
        "LOG_LEVEL": "DEBUG"
      }
    }
  }
}

Configuration

Environment Variables

Performance Considerations

• Pagination: Use appropriate page sizes (10-50 items) to balance response time and data completeness
• Automatic Retry: The server implements intelligent retry logic with:
  • Exponential backoff with jitter to prevent thundering herd
  • Circuit breaker pattern to prevent cascade failures
  • Retry budget to limit resource consumption
  • Respect for Retry-After headers from the API
• Rate Limits: Rate-limited requests (429) are automatically retried with appropriate delays
• Fault Tolerance: Transient failures (network, 502, 503, 504) are handled gracefully
• Caching: Results are not cached. Consider implementing caching for frequently accessed data

Available Tools

1. projects

List all available DeepSource projects.

Parameters: None

Example Response:

[
  {
    "key": "https://api-key@app.deepsource.com",
    "name": "my-python-project"
  }
]

2. project_issues

Get issues from a DeepSource project with filtering and pagination.

Example Response:

{
  "issues": [
    {
      "id": "T2NjdXJyZW5jZTpnZHlqdnlxZ2E=",
      "title": "Avoid using hardcoded credentials",
      "shortcode": "PY-D100",
      "category": "SECURITY",
      "severity": "CRITICAL",
      "file_path": "src/config.py",
      "line_number": 42
    }
  ],
  "totalCount": 15,
  "pageInfo": {
    "hasNextPage": true,
    "endCursor": "YXJyYXljb25uZWN0aW9uOjQ="
  }
}

3. runs

List analysis runs for a project with filtering.

4. run

Get details of a specific analysis run.

5. recent_run_issues

Get issues from the most recent analysis run on a branch.

6. dependency_vulnerabilities

Get security vulnerabilities in project dependencies.

Example Response:

{
  "vulnerabilities": [
    {
      "id": "VUL-001",
      "package": "requests",
      "version": "2.25.0",
      "severity": "HIGH",
      "cve": "CVE-2021-12345",
      "description": "Remote code execution vulnerability"
    }
  ],
  "totalCount": 3
}

7. quality_metrics

Get code quality metrics with optional filtering.

Available Metrics:

• `LCV` - Line Coverage
• `BCV` - Branch Coverage
• `DCV` - Documentation Coverage
• `DDP` - Duplicate Code Percentage
• `SCV` - Statement Coverage
• `TCV` - Total Coverage
• `CMP` - Code Maturity

8. update_metric_threshold

Update the threshold for a quality metric.

9. update_metric_setting

Update metric reporting and enforcement settings.

10. compliance_report

Get security compliance reports.

Available Report Types:

• `OWASP_TOP_10` - Web application security vulnerabilities
• `SANS_TOP_25` - Most dangerous software errors
• `MISRA_C` - Guidelines for safety-critical C code
• `CODE_COVERAGE` - Code coverage report
• `CODE_HEALTH_TREND` - Quality trends over time
• `ISSUE_DISTRIBUTION` - Issue categorization
• `ISSUES_PREVENTED` - Prevented issues count
• `ISSUES_AUTOFIXED` - Auto-fixed issues count

Usage Examples

Monitor Code Quality Trends

Track your project's quality metrics over time:

"Show me the code coverage trend for my main branch"

This combines multiple tools to:

1. Get recent runs for the main branch

2. Retrieve coverage metrics for each run

3. Display the trend

Set Up Quality Gates

Implement quality gates for CI/CD:

"Set up quality gates: 80% line coverage, 0 critical security issues"

This will:

1. Update the line coverage threshold to 80%

2. Configure enforcement for the threshold

3. Check current critical security issues

Investigate Security Vulnerabilities

Comprehensive security analysis:

"Analyze all security vulnerabilities in my project including dependencies"

This performs:

1. Dependency vulnerability scan

2. Code security issue analysis

3. OWASP Top 10 compliance check

4. Prioritized remediation suggestions

Code Review Assistance

Get AI-powered code review insights:

"What are the most critical issues in the recent commits to feature/new-api?"

This will:

1. Find the most recent run on the branch

2. Filter for critical and high severity issues

3. Group by file and issue type

4. Suggest fixes

Team Productivity Metrics

Track team code quality metrics:

"Show me code quality metrics across all our Python projects"

This aggregates:

1. Coverage metrics per project

2. Issue counts by severity

3. Trends over the last month

4. Team performance insights

Architecture

The DeepSource MCP Server uses modern TypeScript patterns for maintainability and type safety.

Key Components

┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐
│  Claude/AI      │────▶│   MCP Server     │────▶│  DeepSource API │
│  Assistant      │◀────│  (TypeScript)    │◀────│   (GraphQL)     │
└─────────────────┘     └──────────────────┘     └─────────────────┘

1. MCP Server Integration (`src/index.ts`)

2. DeepSource Client (`src/deepsource.ts`)

3. Type System (`src/types/`)

Type Safety Features

Branded Types

// Prevent mixing different ID types
type ProjectKey = string & { readonly __brand: 'ProjectKey' };
type RunId = string & { readonly __brand: 'RunId' };

Discriminated Unions

type RunState =
  | { status: 'PENDING'; queuePosition?: number }
  | { status: 'SUCCESS'; finishedAt: string }
  | { status: 'FAILURE'; error?: { message: string } };

Development

Prerequisites

• Node.js 22.19.0 or higher
• pnpm 10.15.1 or higher
• Docker (optional, for container builds)

Setup

# Clone the repository
git clone https://github.com/sapientpants/deepsource-mcp-server.git
cd deepsource-mcp-server

# Install dependencies
pnpm install

# Build the project
pnpm run build

# Run tests
pnpm test

Development Commands

Note: MCP servers communicate via stdio and cannot be run standalone. Use `pnpm run inspect` for interactive debugging.

Troubleshooting & FAQ

Common Issues

Authentication Error

Error: Invalid API key or unauthorized access

Solution: Verify your `DEEPSOURCE_API_KEY` is correct and has necessary permissions.

No Projects Found

Error: No projects found

Solution: Ensure your API key has access to at least one project in DeepSource.

Rate Limit Exceeded

Error: API rate limit exceeded

Solution: The server implements automatic retry. Wait a moment or reduce request frequency.

Pagination Cursor Invalid

Error: Invalid cursor for pagination

Solution: Cursors expire. Start a new pagination sequence from the beginning.

FAQ

Q: Which DeepSource plan do I need?

A: The MCP server works with all DeepSource plans. Some features like security compliance reports may require specific plan features.

Q: Can I use this with self-hosted DeepSource?

A: Yes, configure the API endpoint in your environment variables (feature coming in v1.3.0).

Q: How do I debug issues?

A: Enable debug logging by setting `LOG_LEVEL=DEBUG` and check the log file specified in `LOG_FILE`.

Q: Is my API key secure?

A: The API key is only stored in your local Claude Desktop configuration and is never transmitted except to DeepSource's API.

Q: Can I contribute custom tools?

A: Yes! See the Contributing section for guidelines.

Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Workflow

1. Fork the repository

2. Create a feature branch (`git checkout -b feature/amazing-feature`)

3. Make your changes

4. Run tests (`pnpm test`)

5. Commit your changes using conventional commits (see below)

6. Push to the branch (`git push origin feature/amazing-feature`)

7. Open a Pull Request

Commit Message Convention

This project uses Conventional Commits to ensure consistent commit messages. Commits are validated using commitlint.

Format

[optional scope]: 

[optional body]

[optional footer(s)]

Types

• `feat`: New feature
• `fix`: Bug fix
• `docs`: Documentation only changes
• `style`: Changes that don't affect code meaning (formatting, etc)
• `refactor`: Code change that neither fixes a bug nor adds a feature
• `perf`: Performance improvements
• `test`: Adding missing tests or correcting existing tests
• `build`: Changes that affect the build system or dependencies
• `ci`: Changes to CI configuration files and scripts
• `chore`: Other changes that don't modify src or test files
• `revert`: Reverts a previous commit

Examples

# Feature
git commit -m "feat: add support for filtering issues by severity"

# Bug fix with scope
git commit -m "fix(api): handle null response from DeepSource API"

# Breaking change
git commit -m "feat!: change API response format

BREAKING CHANGE: Response format now uses camelCase instead of snake_case"

Code Standards

• Follow TypeScript best practices
• Maintain test coverage above 80%
• Use meaningful commit messages
• Update documentation for new features

License

MIT - see LICENSE file for details.

External Resources

• Model Context Protocol Specification
• DeepSource Documentation
• DeepSource API Reference

Made with ❤️ by the DeepSource MCP Server community