@whatsdownai/mcp-server

MCP (Model Context Protocol) server for WhatsDown AI - Monitor AI service status and report issues directly from your AI coding agents.

Features

• Check Service Status: Query the operational status of AI services before making API calls
• Report Issues: Report errors, timeouts, and outages to help the community detect issues faster
• List Services: Discover all monitored services and their categories

Installation

Using npx (recommended)

npx @whatsdownai/mcp-server

Global installation

npm install -g @whatsdownai/mcp-server
whatsdownai-mcp

Configuration

Environment Variables

| Variable | Required | Description | |----------|----------|-------------| | WHATSDOWNAI_API_KEY | Yes (for reporting) | Your WhatsDown AI API key for submitting reports | | WHATSDOWNAI_BASE_URL | No | Custom WhatsDown AI instance URL (default: https://api.whatsdown.ai) |

Get your API key at whatsdown.ai/integrate

Usage with AI Agents

Claude Code / Claude Desktop

Add to your claude_desktop_config.json or MCP settings:

{
  "mcpServers": {
    "whatsdownai": {
      "command": "npx",
      "args": ["-y", "@whatsdownai/mcp-server"],
      "env": {
        "WHATSDOWNAI_API_KEY": "your_api_key_here"
      }
    }
  }
}

Cursor

Add to your MCP configuration:

{
  "whatsdownai": {
    "command": "npx",
    "args": ["-y", "@whatsdownai/mcp-server"],
    "env": {
      "WHATSDOWNAI_API_KEY": "your_api_key_here"
    }
  }
}

Cline (VS Code)

Add to Cline's MCP server settings:

{
  "whatsdownai": {
    "command": "npx",
    "args": ["-y", "@whatsdownai/mcp-server"],
    "env": {
      "WHATSDOWNAI_API_KEY": "your_api_key_here"
    }
  }
}

Available Tools

check_all_services

Get the operational status of all monitored AI services.

Parameters:

• category (optional): Filter by category
  • model_providers - OpenAI, Anthropic, Mistral, etc.
  • mcp_servers - GitHub MCP, Slack MCP, etc.
  • agent_frameworks - LangChain, CrewAI, etc.
  • skills_marketplace - Composio, etc.
  • agent_providers - OpenAI Agents, Vertex AI, etc.

Example response:

{
  "summary": {
    "total": 25,
    "operational": 23,
    "degraded": 1,
    "partial_outage": 1,
    "major_outage": 0
  },
  "issues": [
    {
      "service": "OpenAI",
      "slug": "openai",
      "status": "degraded",
      "category": "model_providers"
    }
  ]
}

check_service_status

Check a specific service's operational status.

Parameters:

• service (required): Service slug (e.g., "openai", "anthropic", "github-mcp")

Example response:

{
  "service": "OpenAI",
  "slug": "openai",
  "status": "operational",
  "category": "model_providers",
  "operational": true,
  "recommendation": "Service is operational. Safe to proceed with API calls."
}

report_service_issue

Report an issue with a service. Requires WHATSDOWNAI_API_KEY.

Parameters:

• service (required): Service slug experiencing issues
• error_type (required): Type of error
  • backend_down - Service completely unreachable
  • timeout - Request timed out
  • rate_limited - Hit rate limits
  • auth_failure - Authentication failed
  • mcp_failure - MCP protocol error
  • invalid_response - Malformed response
  • other - Other issues
• description (required): Detailed error description
• severity (optional): low, medium, high, critical (default: medium)

Example response:

{
  "success": true,
  "report_id": "abc123",
  "message": "Report submitted for OpenAI",
  "status_updated": false
}

list_services

List all monitored services with their slugs.

Parameters:

• category (optional): Filter by category

Resources

The server also exposes MCP resources:

• agentup://services/status - All service statuses as JSON
• agentup://services/issues - Only services with current issues

Best Practices

1. Check before calling: Always check service status before making API calls to external services
2. Report issues: When you encounter errors, report them to help the community
3. Include details: When reporting, include error messages and context
4. Use appropriate severity: Reserve critical for complete service failures

Example Workflow

Agent: I need to call the OpenAI API to generate some code.

1. First, let me check if OpenAI is operational...
   [calls check_service_status with service="openai"]

2. OpenAI shows "operational" - proceeding with API call...

3. [API call fails with timeout]

4. Let me report this issue...
   [calls report_service_issue with service="openai",
    error_type="timeout", description="API timeout after 30s on chat completion"]

5. Report submitted. Trying alternative service...
   [calls check_service_status with service="anthropic"]

License

MIT