Error Tracker MCP Server

Model Context Protocol (MCP) server for Error Tracker - Query and resolve errors directly from your terminal or IDE using AI assistants like Claude Desktop, Warp, and Cursor.

⚡ Quick Start

• Go to the dashboard: https://error-tracker-web.vercel.app/
• Create your organization and then a project
• In the project: Settings → API Keys → generate your token (API Key)
• Configure your MCP client:

{
  "error-tracker": {
    "command": "npx",
    "args": ["-y", "@masmuch/error-tracker-mcp"],
    "env": {
      "ERROR_TRACKER_API_KEY": "et_your_api_key"
    }
  }
}

Note: You don’t need to set ERROR_TRACKER_API_URL for the hosted dashboard; it defaults automatically to the cloud service.

You can now list projects, view errors, and mark them as resolved from your assistant.

Features

• 🔍 List and search errors with powerful filters (status, severity, platform, environment)
• 📊 Get error statistics and analytics
• 🔎 View complete error details including stack traces, context, and metadata
• 🔧 Resolve errors and update status directly from your AI assistant
• 🔗 Find similar errors using fingerprint matching
• 🏢 Multi-tenant security with organization-based access control
• 📝 Audit logging for compliance and tracking

Installation

Option 1: Install from NPM (Recommended)

npm install -g @masmuch/error-tracker-mcp

Option 2: Run with npx (No installation)

npx @masmuch/error-tracker-mcp

Option 3: Build from source

cd apps/mcp-server
pnpm install
pnpm build

Configuration

1. Get your API Key

1. Go to the dashboard: https://error-tracker-web.vercel.app/
2. Sign in and create your organization
3. Create a project inside the organization
4. Open the project → Settings → API Keys and click "Create"
5. Copy the generated key (starts with et_) and save it securely (it's only shown once)

2. Configure your MCP Client

Claude Desktop

Edit ~/.config/claude/config.json (Mac/Linux) or %APPDATA%\Claude\config.json (Windows):

{
  "mcpServers": {
    "error-tracker": {
      "command": "npx",
      "args": ["-y", "@masmuch/error-tracker-mcp"],
      "env": {
        "ERROR_TRACKER_API_KEY": "et_your_api_key_here"
      }
    }
  }
}

Note: ERROR_TRACKER_API_URL is optional and defaults to https://error-tracker-web.vercel.app. Only specify it if you're using a self-hosted instance.

Warp AI

Add to your Warp MCP configuration:

{
  "error-tracker": {
    "command": "npx",
    "args": ["-y", "@masmuch/error-tracker-mcp"],
    "env": {
      "ERROR_TRACKER_API_KEY": "et_your_api_key_here"
    }
  }
}

Cursor

Add to your Cursor MCP settings:

{
  "mcpServers": {
    "error-tracker": {
      "command": "npx",
      "args": ["-y", "@masmuch/error-tracker-mcp"],
      "env": {
        "ERROR_TRACKER_API_KEY": "et_your_api_key_here"
      }
    }
  }
}

3. Restart your client

After editing the configuration, restart Claude Desktop, Warp, or Cursor for the changes to take effect.

Available Tools

1. error-tracker_list_errors

List errors with powerful filtering options.

Parameters:

• projectId (optional): Filter by specific project
• status (optional): UNRESOLVED | IN_PROGRESS | RESOLVED | IGNORED
• severity (optional): CRITICAL | HIGH | MEDIUM | LOW | INFO
• environment (optional): production, staging, development, etc.
• platform (optional): VERCEL, AWS_LAMBDA, NODEJS, BROWSER, etc.
• limit (default: 20): Number of errors to return (1-100)
• offset (default: 0): Pagination offset
• orderBy (default: lastSeen): lastSeen | firstSeen | occurrences
• order (default: desc): asc | desc

Example usage:

"Show me the 10 most recent critical errors in production"

2. error-tracker_get_error_details

Get complete details of a specific error.

Parameters:

• errorId (required): The error ID

Example usage:

"Show me the full details of error abc123"

3. error-tracker_search_errors

Search errors by text in message, stack trace, type, or filename.

Parameters:

• query (required): Search query
• projectId (optional): Filter by project
• status (optional): Filter by status
• severity (optional): Filter by severity
• limit (default: 20): Number of results

Example usage:

"Search for errors containing 'null pointer'"

4. error-tracker_get_error_stats

Get error statistics and analytics.

Parameters:

• projectId (optional): Get stats for specific project
• days (default: 7): Number of days to include (1-90)

Example usage:

"Show me error statistics for the last 30 days"

5. error-tracker_list_projects

List all accessible projects with error counts.

Parameters:

• includeErrorCount (default: true): Include count of active errors

Example usage:

"What projects do I have access to?"

6. error-tracker_resolve_error

Update error status and add resolution notes.

Parameters:

• errorId (required): The error ID
• status (required): UNRESOLVED | IN_PROGRESS | RESOLVED | IGNORED
• resolution (required for RESOLVED): Notes about the resolution

Example usage:

"Mark error abc123 as resolved with note 'Fixed null check in user validation'"

7. error-tracker_get_similar_errors

Find errors with similar fingerprints (same root cause).

Parameters:

• errorId (required): Reference error ID
• limit (default: 10): Number of similar errors (1-50)

Example usage:

"Find errors similar to abc123"

Usage Examples

Example 1: Find and fix critical errors

User: What are the critical errors in my web-app project?

AI uses: error-tracker_list_projects → gets project ID
         error-tracker_list_errors (projectId, severity=CRITICAL)

AI: You have 3 critical errors:
1. TypeError: Cannot read property 'name' of undefined
   - Last seen: 2 hours ago
   - Occurrences: 42
   ...

Example 2: Deep dive into an error

User: Show me the full stack trace for error abc123

AI uses: error-tracker_get_error_details (errorId: abc123)

AI: Here's the complete error:
- Message: Cannot read property 'name' of undefined
- File: src/components/UserProfile.tsx:45
- Stack trace: [full trace]
- User context: { id: "user_123", email: "[email protected]" }
- Request: { url: "/api/users/profile", method: "GET" }
...

Example 3: Resolve an error

User: Mark error abc123 as resolved, I fixed the null check

AI uses: error-tracker_resolve_error (
           errorId: abc123,
           status: RESOLVED,
           resolution: "Fixed null check in UserProfile component"
         )

AI: ✅ Error abc123 has been marked as resolved.
    The audit log has been updated.

Example 4: Find patterns

User: Are there other errors like abc123?

AI uses: error-tracker_get_similar_errors (errorId: abc123)

AI: Found 5 similar errors with the same fingerprint:
1. Error def456 - Last seen 1 hour ago
2. Error ghi789 - Last seen 3 hours ago
...

All seem to be caused by the same null pointer issue.

Security

• API Key Authentication: All requests require a valid API key
• Organization Isolation: You can only access errors from your organization's projects
• Project-Level Keys: Optionally create API keys limited to specific projects
• Audit Logging: All actions are logged for compliance
• Rate Limiting: Protection against abuse (configurable)

Troubleshooting

Error: "API key is required"

Make sure ERROR_TRACKER_API_KEY is set in your MCP client configuration.

Error: "Failed to connect to API"

Check your internet connection. The default API URL is https://error-tracker-web.vercel.app. If using a self-hosted instance, set ERROR_TRACKER_API_URL in your configuration.

Error: "API key not found"

Your API key may be invalid or deleted. Generate a new one from the dashboard.

Error: "API key is disabled"

Your API key has been disabled. Enable it in the dashboard or create a new one.

Error: "Access denied to this project"

Your API key doesn't have access to the requested project. Check your permissions in the dashboard.

MCP server not showing up

1. Check that your configuration file is valid JSON/YAML
2. Restart your MCP client completely
3. Check the client logs for errors
4. Try running the server manually: npx @masmuch/error-tracker-mcp

Testing the server manually

You can test the server using the MCP Inspector:

npx @modelcontextprotocol/inspector npx @masmuch/error-tracker-mcp

This will open a web interface where you can test all tools interactively.

Development

Building

pnpm build

Running in development

pnpm dev

Testing with MCP Inspector

pnpm build
npx @modelcontextprotocol/inspector node dist/index.js

Links

• Error Tracker Dashboard
• SDK Documentation
• MCP Protocol Specification

License

MIT

Support

For issues or questions:

• Open an issue on GitHub
• Contact support through the dashboard
• Email: [email protected]