Skip to main content
Glama
tjdam007

Firebase Crashlytics MCP Server

by tjdam007
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

🔥 Firebase Crashlytics MCP Server

License: MIT Node.js MCP

A clean, simple Model Context Protocol (MCP) server that integrates Firebase Crashlytics with BigQuery for AI-powered crash analysis. Built for use with Claude Code and other MCP-compatible AI assistants.

✨ Features

  • 🔍 Simple Tools: Clean, single-responsibility tools for crash analysis

  • 📱 App Discovery: Automatically discover all apps in your Firebase project

  • 💥 Fatal Crash Analysis: Get detailed fatal crashes with stack traces

  • 🐛 ANR Issue Detection: Analyze Application Not Responding issues

  • 🔧 Easy Setup: Simple configuration with environment variables

  • 🤖 AI-Optimized: Structured output perfect for AI analysis and debugging

Related MCP server: Google Tasks MCP Server

🚀 Quick Start

Prerequisites

  • Node.js 18+

  • Firebase project with Crashlytics enabled

  • BigQuery export enabled for Crashlytics

  • Google Cloud service account with BigQuery access

Installation

  1. Clone the repository:

git clone https://github.com/your-username/mcp-crashlytics-server.git
cd mcp-crashlytics-server

  1. Install dependencies:

npm install

  1. Build the project:

npm run build

Configuration

  1. Create environment file:

cp .env.example .env

  1. Configure your environment variables:

# Google Cloud Service Account (choose one method)
GOOGLE_SERVICE_ACCOUNT_KEY=/path/to/your-service-account.json
# OR base64-encoded: GOOGLE_SERVICE_ACCOUNT_KEY=eyJhbGciOiJIUzI1...

# Your Firebase project settings
BIGQUERY_PROJECT_ID=your-firebase-project-id
BIGQUERY_DATASET_ID=firebase_crashlytics

# Optional: Default limit for crash queries
DEFAULT_CRASH_LIMIT=10

Setup Firebase & BigQuery

  1. Enable Crashlytics BigQuery Export:

    • Go to Firebase Console → Project Settings → Integrations

    • Enable BigQuery integration for Crashlytics

    • Data will be exported to firebase_crashlytics dataset

  2. Create Service Account:

    • Go to Google Cloud Console

    • Navigate to IAM & Admin → Service Accounts

    • Create service account with these roles:

      • BigQuery Data Viewer

      • BigQuery Job User

    • Download JSON key file

Register with AI Agents

Claude Code

# Register the MCP server
claude mcp add crashlytics -- node /path/to/mcp-crashlytics-server/dist/index.js

# Verify it's working
claude mcp list

Cursor

Add to your .cursorrules or cursor settings:

{
  "mcpServers": {
    "crashlytics": {
      "command": "node",
      "args": ["/path/to/mcp-crashlytics-server/dist/index.js"],
      "env": {
        "GOOGLE_SERVICE_ACCOUNT_KEY": "/path/to/your-service-account.json",
        "BIGQUERY_PROJECT_ID": "your-firebase-project-id",
        "BIGQUERY_DATASET_ID": "firebase_crashlytics"
      }
    }
  }
}

Windsurf

Add to your Windsurf settings or .windsurfrules:

{
  "mcp": {
    "servers": {
      "crashlytics": {
        "command": "node",
        "args": ["/path/to/mcp-crashlytics-server/dist/index.js"],
        "env": {
          "GOOGLE_SERVICE_ACCOUNT_KEY": "/path/to/your-service-account.json",
          "BIGQUERY_PROJECT_ID": "your-firebase-project-id", 
          "BIGQUERY_DATASET_ID": "firebase_crashlytics"
        }
      }
    }
  }
}

Other MCP-Compatible IDEs

For other MCP-compatible tools, add this server configuration:

  • Command: node

  • Args: ["/path/to/mcp-crashlytics-server/dist/index.js"]

  • Environment Variables: Same as above (.env file values)

🛠️ Available Tools

1. list_available_apps

Discover all apps in your Firebase Crashlytics dataset.

  • Parameters: None

  • Returns: List of apps with crash counts

// Usage in Claude Code/Cursor
"What apps are available in my crashlytics data?"

2. get_fatal_crashes

Get fatal crashes for a specific app.

  • Parameters:

    • app_package (required): App package name (e.g., "com.example.myapp")

    • limit (optional): Number of crashes (1-50, default: 10)

// Usage in Claude Code/Cursor  
"Get the 10 most recent fatal crashes for com.example.myapp"

3. get_anr_issues

Get ANR (Application Not Responding) issues for a specific app.

  • Parameters:

    • app_package (required): App package name

    • limit (optional): Number of issues (1-50, default: 10)

// Usage in Claude Code/Cursor
"Show me ANR issues for com.example.myapp, limit to 5"

📋 Example Usage

Once set up, you can ask Claude Code natural questions like:

# Discover your apps
"What apps do I have crash data for?"

# Get fatal crashes
"Show me the latest fatal crashes for my Android app"
"Get 5 fatal crashes for com.mycompany.myapp"

# Analyze ANR issues  
"What ANR issues does my app have?"
"Show ANR problems for com.mycompany.myapp from the last week"

🏗️ Project Structure

mcp-crashlytics-server/
├── src/
│   ├── index.ts              # Main MCP server
│   ├── bigquery-client.ts    # BigQuery integration  
│   ├── crash-processor.ts    # Crash data processing
│   ├── impact-analyzer.ts    # Impact analysis
│   └── types.ts             # TypeScript definitions
├── dist/                    # Built JavaScript files
├── package.json
├── tsconfig.json
├── .env.example            # Environment template
└── README.md

🔧 Development

Running in Development Mode

npm run dev

Building

npm run build

Linting

npm run lint

Testing

npm test

🐛 Troubleshooting

Common Issues

1. "No MCP servers configured"

  • Make sure to register the server: claude mcp add crashlytics -- node /path/to/dist/index.js

  • Restart Claude Code/Cursor IDE

2. "Authentication Failed"

  • Verify service account JSON file path is correct

  • Ensure service account has BigQuery permissions

  • Check project ID matches your Firebase project

3. "No crashes returned"

  • Confirm Crashlytics BigQuery export is enabled

  • Wait 24-48 hours for initial data export

  • Verify your app is actually sending crash data

4. "Table not found"

  • Check that BIGQUERY_DATASET_ID is correct (usually firebase_crashlytics)

  • Ensure crashes exist in your Firebase project

  • Verify BigQuery export is working in Firebase Console

Debug Mode

Enable verbose logging:

DEBUG=mcp-crashlytics-server npm start

🤝 Contributing

  1. Fork the repository

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

  3. Make your changes

  4. Add tests for new functionality

  5. Commit your changes (git commit -m 'Add amazing feature')

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

  7. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🔒 Security Considerations

  • ⚠️ Never expose service account credentials in logs or version control

  • 🔐 Use least-privilege permissions for service accounts

  • 🔑 Store credentials securely using environment variables

  • 🚫 Add .env to .gitignore (already included)

📚 Resources

💡 Need Help?

Made with ❤️ for the MCP community

Available Tools

5 tools
get_anr_issuesB

Get ANR (Application Not Responding) issues for a specific app package name

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNoNumber of ANR issues to fetch (default: 10, max: 50)
app_packageYesApp package name (e.g., com.example.myapp)

TDQS

B3.4/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description must fully disclose behavioral traits. It only states it 'gets' issues, implying a read operation, but fails to mention any side effects, authentication needs, rate limits, or output structure. The lack of output schema further increases the burden.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single sentence, front-loading the core functionality. It contains no redundant or unnecessary words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given no output schema and no annotations, the description lacks details about return values, pagination, or error conditions. For a tool with simple parameters but part of a larger suite, more context would improve usability.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the baseline is 3. The description does not add additional semantic value beyond the schema, as it only repeats the app package name concept.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the action ('Get') and the resource ('ANR issues') with the required parameter ('for a specific app package name'). It uniquely identifies the tool's purpose compared to siblings like 'analyze_crash_trends' or 'get_crash_details'.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description does not provide explicit guidance on when to use this tool versus alternatives. While the tool name and sibling list imply it is for ANR-specific data, no when-to-use or when-not-to-use information is given.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_crash_detailsC

Get detailed information for a specific crash

ParametersJSON Schema
NameRequiredDescriptionDefault
crash_idYesUnique crash identifier

TDQS

C2.9/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided; the description only says 'detailed information' without specifying what data is included (e.g., stack traces, timestamps). No mention of access permissions or side effects.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, clear sentence with no wasted words. It is appropriately sized for a simple one-parameter tool.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

No output schema is provided, and the description omits what 'detailed information' includes. Given the lack of annotations and output schema, the description is insufficiently complete for an agent to understand the full behavior.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The single parameter 'crash_id' is documented in the schema with description 'Unique crash identifier'. The description adds no further semantic value, so baseline score of 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description 'Get detailed information for a specific crash' clearly states the action and resource, but does not explicitly distinguish from sibling 'get_fatal_crashes' which returns a list.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool vs alternatives. Users must infer from context that it is for a single crash identified by an ID, while siblings like 'analyze_crash_trends' serve different purposes.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_fatal_crashesB

Get fatal crashes for a specific app package name

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNoNumber of fatal crashes to fetch (default: 10, max: 50)
app_packageYesApp package name (e.g., com.example.myapp)

TDQS

B3.1/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the burden but only states the basic operation. It does not disclose return type, pagination, read-only nature, or authentication requirements.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single clear sentence with no unnecessary words. While short, it is appropriately sized for a simple tool, though it could include more detail without losing conciseness.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Without an output schema, the description should explain what is returned. It does not mention the structure of fatal crashes, limit behavior, or ordering. Sibling tool contexts hint at subtypes but are not used.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100% with descriptions for both parameters. The description adds no extra meaning beyond what is already in the schema, so baseline score is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the verb 'Get' and resource 'fatal crashes', specifying it is for a specific app package name. This distinguishes it from siblings like get_anr_issues or get_crash_details.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance is provided on when to use this tool versus alternatives like analyze_crash_trends or get_crash_details. The description does not mention exclusions or preferred contexts.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

list_available_appsB

List all available apps in the Firebase Crashlytics dataset

ParametersJSON Schema
NameRequiredDescriptionDefault

No parameters

TDQS

B3.3/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description fully bears the burden of disclosing behavior. It does not mention authentication requirements, data scope, or any side effects. The description is too minimal to inform an agent about important behavioral traits.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, concise sentence that front-loads the action and resource. No unnecessary words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The description is adequate for a simple parameterless list tool, but it lacks information about return values or prerequisites. Given no output schema, it could benefit from stating what fields are returned (e.g., app IDs, names).

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema has no parameters (100% coverage by default). The description does not add any parameter-specific meaning beyond the empty schema, so baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the action (list) and the resource (available apps in Firebase Crashlytics dataset). It is specific and distinct from sibling tools, which focus on crash analysis and details.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool versus alternatives. While it is a list operation that likely precedes other tools, this is not explicitly stated.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 5 tool updatesv1.0.0
    • First observedanalyze_crash_trends
    • First observedget_anr_issues
    • First observedget_crash_details
    • First observedget_fatal_crashes
    • First observedlist_available_apps
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.2/5.0

Scored across 5 tools

Disambiguation5/5

Each tool targets a distinct aspect of crash analysis: apps, trends, fatal crashes, ANR issues, and crash details. No functional overlap.

Naming Consistency5/5

All five tools follow a consistent verb_noun pattern in snake_case (e.g., list_available_apps, get_fatal_crashes), making naming predictable and clear.

Tool Count4/5

Five tools is a reasonable size for a focused Crashlytics server, covering core read operations without being too bare or excessive.

Completeness3/5

The set covers trends, fatal crashes, ANR issues, and crash details but lacks non-fatal crashes and any mutation capabilities, which are notable gaps for a crash analysis server.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers