Prismix Status
Demandez à Claude si un fournisseur d'IA est en panne — un serveur MCP hébergé avec statut en temps réel, historique des incidents et disponibilité sur 30 jours pour plus de 75 services d'IA, sans authentification.
Documentation
Prismix Status MCP
Ask Claude "Is OpenAI down?" and get a live answer.
A Model Context Protocol (MCP) server exposing real-time operational status for 75+ AI services — OpenAI, Anthropic, Cursor, GitHub Copilot, Gemini, Mistral, Perplexity, DeepSeek, Hugging Face, and more. Powered by Prismix, which monitors each provider's official status feed around the clock.
- Hosted endpoint, zero setup:
https://prismix.dev/api/v1/mcp - No API key, no signup, no rate-limit hoops
- Streamable HTTP transport, MCP protocol
2024-11-05 - CORS-enabled; works from Claude Desktop, Claude Code, Cursor, or any MCP client
- This repo is the server's source — also deployable as your own Cloudflare Worker
Quick start
Claude Desktop
Add to claude_desktop_config.json (Settings → Developer → Edit Config):
{
"mcpServers": {
"prismix-status": {
"url": "https://prismix.dev/api/v1/mcp"
}
}
}
Restart Claude Desktop. Done — ask "Which AI services are degraded right now?"
Claude Code
claude mcp add --transport http prismix-status https://prismix.dev/api/v1/mcp
Cursor
Add to .cursor/mcp.json in your project (or ~/.cursor/mcp.json globally):
{
"mcpServers": {
"prismix-status": {
"url": "https://prismix.dev/api/v1/mcp"
}
}
}
Tools
check_ai_status
Current status for one service (or all of them).
| Argument | Type | Description |
|---|---|---|
service | string (optional) | Service ID or partial name, case-insensitive — "openai", "anthropic", "cursor". Omit to check everything. |
Returns per service: status indicator (Operational / Degraded / Partial outage / Major outage), active incident count, 30-day uptime %, response latency, and a link to the full incident history.
Things you can ask:
- "Is Anthropic down right now?"
- "What's Cursor's 30-day uptime?"
- "Check the status of every AI service I might be using"
get_incident_history
Recent incident history (last 30 days) for one service — past outages and degradations with impact level, start time, and resolution/duration, plus the 30-day uptime %.
| Argument | Type | Description |
|---|---|---|
service | string (required) | Service ID or partial name — "anthropic", "openai", "cursor". |
Things you can ask:
- "Has Anthropic been reliable this month?"
- "How often does Cursor go down?"
- "Show me OpenAI's recent incidents"
list_ai_services
Enumerates all 75+ tracked services with ID, name, category, and current indicator. Useful for discovering exact service IDs.
Protocol details
- JSON-RPC 2.0 over HTTP POST (Streamable HTTP transport)
- Protocol version
2024-11-05 - Supported methods:
initialize,ping,tools/list,tools/call - Notifications (requests without
id) are accepted with202
Try it from the command line:
curl -s https://prismix.dev/api/v1/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"check_ai_status","arguments":{"service":"openai"}}}'
Deploy your own
The worker sources data from Prismix's free public REST API (GET https://prismix.dev/api/v1/statuses — no auth, CORS-enabled), so a self-hosted copy needs zero secrets:
npm install
npx wrangler deploy
Your worker URL then speaks the same MCP protocol as the hosted endpoint.
Data
Status data refreshes every ~5 minutes from each provider's official status page or a direct HTTP probe where no status page exists. The full dashboard, incident history, per-service uptime, and embeddable badges live at prismix.dev. Email alerts on status changes are free with an account.