WhatsDownAI

Integrate Your Agent

Configure your AI agents to automatically report issues when services fail

Quick Start
Get your agent reporting issues in under 5 minutes
1

Get an API Key

Sign in and generate your API key below

2

Connect to MCP Server

Add our hosted MCP server to Claude Code, Cursor, or your IDE

3

Check Status & Report

Agents check service status and report failures

Your API Keys

API Keys Required

Create a free account to generate API keys for your agents. Read the documentation below to learn how to integrate.

Sign In to Get API Keys

Integration Guide

Hosted MCP Server (Recommended)
Connect directly to our hosted MCP server - no local installation required
Live

Connect your agent directly to our hosted MCP server. No local installation or authentication required for status checks - works with Claude Code, Cursor, Windsurf, VS Code, and more.

Authentication (Optional)

API key authentication is optional. All status-reading tools work without auth. If you want report_issue reports to be associated with your account, pass your API key via the Authorization: Bearer YOUR_KEY header.

Claude Code (CLI)

Add the MCP server (no authentication required):

claude mcp add whatsdownai \
  -- npx -y mcp-remote https://whatsdown.ai/api/mcp/sse

For project-specific config, add --scope project after add

Claude Code / Desktop (.mcp.json)

Add to your .mcp.json or claude_desktop_config.json:

{
  "mcpServers": {
    "whatsdownai": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://whatsdown.ai/api/mcp/sse"]
    }
  }
}

Cursor / Windsurf / VS Code

For editors supporting Streamable HTTP transport:

{
  "mcpServers": {
    "whatsdownai": {
      "url": "https://whatsdown.ai/api/mcp/mcp"
    }
  }
}

With Authentication (Optional)

To link issue reports to your account, add the header flag:

claude mcp add whatsdownai \
  -- npx -y mcp-remote https://whatsdown.ai/api/mcp/sse \
  --header "Authorization:Bearer YOUR_API_KEY"

Available Tools

get_system_status

Get overall system status with counts of operational, degraded, and down services

list_services

List all monitored AI services with current status, filterable by category

get_service_status

Get detailed status for a specific service including active incidents

report_issue

Report issues when API calls fail to help crowdsource outage detection

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

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

2. Response: { "status": "operational", "recommendation": "Safe to proceed" }
   OpenAI is operational - proceeding with API call...

3. [API call fails with timeout after 30s]

4. Let me report this issue to help the community...
   [calls report_issue with:
    service="openai", 
    type="timeout", 
    description="Chat completion API timeout after 30s"]

5. Report submitted. Let me check if Anthropic is available as fallback...
   [calls check_status with service="anthropic"]

6. Anthropic is operational - switching to Claude for this task.

Install via Skills.sh

For agents that support Vercel Skills, install WhatsDownAI with a single command

Quick Install
Add WhatsDownAI monitoring to any agent that supports skills

Install the main skill

npx skills add whatsdownai/whatsdownai

Or install individual skills

Status checking only:

npx skills add whatsdownai/whatsdownai --skill check-ai-service-status

Issue reporting only:

npx skills add whatsdownai/whatsdownai --skill report-ai-service-issue

Available Skills

whatsdownai
Complete skill

Both status checking and issue reporting in one skill

check-ai-service-status
Pre-flight checks

Check if services are operational before making API calls

report-ai-service-issue
Error reporting

Report issues when you encounter errors with AI services

Setup for Popular Coding Agents

Instructions for integrating WhatsDownAI with the most popular AI coding assistants

C
Claude Code (Anthropic)
Integrate WhatsDownAI with Claude Code CLI

Recommended: Use the MCP server for full status checking + reporting capabilities

1. Add MCP Server (Recommended)

Add WhatsDownAI to your Claude Code MCP settings for status checking and issue reporting:

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

2. Add to CLAUDE.md

Add instructions to your project's CLAUDE.md:

# WhatsDownAI Integration

You have access to WhatsDownAI tools via MCP for monitoring AI service health:

## Before Making External API Calls
Use check_service_status to verify services are operational before calling them.
Example: check_service_status(service="openai")

## When Encountering Errors
Report issues using report_service_issue to help detect outages early.
Example: report_service_issue(service="openai", error_type="timeout", description="API timeout after 30s")

## Available Services
Model providers: openai, anthropic, mistral, cohere, groq, deepseek, perplexity
MCP servers: github-mcp, slack-mcp, stripe-mcp, cloudflare-mcp, neon-mcp

Best Practices

Avoid Duplicate Reports

Implement rate limiting on your side - don't report the same issue more than once per minute. Our API will deduplicate, but it's better to handle client-side.

Include Metadata

Add your agent name, version, and any relevant context to metadata. This helps identify patterns across different agent implementations.

Check Status First

Before making API calls, optionally check if a service is already experiencing issues. This can help you fail fast or use fallbacks.

Secure Your API Key

Store your API key in environment variables, never in code. Rotate keys periodically and revoke any that may have been exposed.