Outage data for humans, apps and AI agents.
A free JSON API, signed webhooks and a remote MCP server. Check whether your dependencies are up before you run, and get pushed an alert when they aren't.
Status API
No key needed. Responses are JSON and cached for up to 60 seconds. Each service is checked about every 11 minutes.
GET /api/v1/services | All monitored services and categories |
GET /api/v1/status | Every service's verdict. Add ?slugs=a,b,c for a readiness check of your stack |
GET /api/v1/status/{slug} | One service, with active official incidents |
GET /api/v1/incidents/{slug} | Recent official incidents and verdict changes |
GET /api/v1/services/{slug}/history?days=7|30 | Uptime, incident count, and p50/p95 of our HTTP probe |
GET /api/v1/ai/health | AI API providers, current status, and healthy fallbacks |
GET /api/v1/ai/fallback?provider={slug} | Best healthy alternatives for one AI provider |
POST /api/v1/reports | Agent/client issue report (5xx, timeout, rate_limit, …) |
curl "https://whatsdownnow.com/api/v1/status?slugs=chatgpt,claude,gemini,aws,vercel,supabase,stripe,cloudflare,github"
{
"overall": "degraded",
"ready": false,
"summary": "Supabase: degraded",
"services": [ { "slug": "chatgpt", "verdict": "up", "summary": "...", "user_reports_last_hour": 0, ... } ]
}
Verdicts: up, degraded, down, unknown. They combine each service's official status page with our own HTTP checks and can lag or be wrong. Confirm critical decisions on the official page.
AI provider fallbacks
A board at https://whatsdownnow.com/ai watches the model and inference APIs we already track — OpenAI, Anthropic, Gemini, Mistral, Groq, Perplexity, Cohere, xAI, DeepSeek, Together, Replicate, Hugging Face, OpenRouter and Fireworks — and suggests currently healthy alternatives when one is degraded or down.
curl https://whatsdownnow.com/api/v1/ai/health
curl "https://whatsdownnow.com/api/v1/ai/fallback?provider=chatgpt"
curl "https://whatsdownnow.com/api/v1/services/claude/history?days=7"
Latency in these responses is from our HTTP probe of the provider's public website, API or status page, not model inference latency. We do not make paid model API calls.
From MCP, call get_ai_fallback (omit provider for the whole board). get_status includes the same fallback hints for AI APIs. If your agent hits a 5xx, timeout or rate limit, call report_issue (or POST /api/v1/reports) so a cluster of independent reports can be flagged as an agent-reported spike.
TypeScript, Python and GitHub Action
A tiny zero-dependency TypeScript client, a Python client, and a composite GitHub Action live in the WhatsDownNow repo under /packages. They are not on npm, PyPI or the GitHub Marketplace yet — those listings are coming soon. Until then, copy the files from the repo (if you have access) or use the snippets below.
TypeScript (npm coming soon). Install-from-repo, when you have access:
# npm coming soon. For now, copy packages/js into your project.
import { WhatsDownNow } from "./whatsdownnow";
const wdn = new WhatsDownNow();
const status = await wdn.getStatus("chatgpt");
const health = await wdn.aiHealth();
const fb = await wdn.fallback("claude");
// Verify a signed webhook
wdn.verifyWebhook(rawBody, req.headers.get("x-wdn-signature"), signingSecret);
Python (PyPI coming soon).
from whatsdownnow import WhatsDownNow
wdn = WhatsDownNow()
print(wdn.get_status("chatgpt"))
print(wdn.ai_health())
print(wdn.fallback("openai"))
wdn.verify_webhook(raw_body, signature_header, signing_secret)
GitHub Action (Marketplace coming soon). Until it's listed, call the API from a workflow, or copy packages/github-action into your repo:
# .github/workflows/providers.yml
# Marketplace listing coming soon. Copy packages/github-action, or:
jobs:
providers:
runs-on: ubuntu-latest
steps:
- name: Check AI providers
run: |
curl -fsS -A "WhatsDownNow-Action/1.0" "https://whatsdownnow.com/api/v1/status?slugs=chatgpt,claude,gemini" -o /tmp/wdn.json
python3 -c "import json,sys; d=json.load(open('/tmp/wdn.json')); sys.exit(0 if d.get('ready') else 1)"
Alert stacks
A stack is a list of services plus alert rules and delivery channels. There are no accounts: creating a stack returns a management token, shown once. We store only its hash. Send it as Authorization: Bearer <token>.
curl -X POST https://whatsdownnow.com/api/v1/stacks -H 'content-type: application/json' -d '{
"name": "Acme AI production",
"services": ["chatgpt", "claude", "aws", "stripe",
{"slug": "supabase", "degraded_minutes": 5, "reports_per_hour": 20}],
"defaults": {"on_down": true, "degraded_minutes": 15, "reports_per_hour": null},
"notify_recovery": true,
"webhooks": ["https://example.com/hooks/outages"],
"emails": ["oncall@example.com"]
}'
Rules per service: on_down (alert on a major outage), degraded_minutes (degraded or down for at least N minutes), reports_per_hour (at least N user reports in the last hour). Each incident alerts once, plus once on recovery. Thresholds relax while an alert is open, so they don't flap.
GET /api/v1/stacks/{id} | Stack, live status of each service, channels, recent deliveries |
PATCH /api/v1/stacks/{id} | Change name, services and rules, webhooks, emails, push subscriptions |
DELETE /api/v1/stacks/{id} | Delete the stack and stop all alerts |
POST /api/v1/stacks/{id}/test | Send a test alert to every active channel now |
POST /api/v1/stacks/{id}/rotate-key | Replace the management token and manage link (shown once). The old ones stop working immediately. |
POST /api/v1/stacks/{id}/push | Add a browser PushSubscription (VAPID key: /api/v1/push/vapid-public-key) |
DELETE /api/v1/stacks/{id}/channels/{channelId} | Remove one channel |
Email uses double opt-in: the address gets a confirmation link and receives nothing until it's clicked. Every alert email has a one-click unsubscribe link (List-Unsubscribe and List-Unsubscribe-Post headers).
Webhooks
We POST JSON to your URL. Slack (hooks.slack.com) and Discord webhook URLs get their native message format automatically.
{
"event": "alert.triggered", // or alert.recovered, alert.test
"event_id": "evt_...", "delivery_id": "dl_...", "attempt": 1,
"stack": { "id": "stk_...", "name": "Acme AI production" },
"service": { "slug": "claude", "name": "Claude / Anthropic", "category": "AI",
"status_url": "https://whatsdownnow.com/status/claude", "official_status_page": "https://status.claude.com/" },
"verdict": { "old": "up", "new": "down" },
"reason": "Major outage (Down)",
"why": "Official status: Major outage - API",
"user_reports_last_hour": 12,
"occurred_at": "2026-11-17T22:40:00.000Z",
"checked_at": "2026-11-17T22:35:12.000Z",
"incident_started_at": "2026-11-17T22:35:12.000Z",
"payload_version": 2,
"started_at": "2026-11-17T22:35:12.000Z",
"severity": "critical",
"source": "official",
"component": "API",
"model": null,
"suggested_fallbacks": [
{ "slug": "openrouter", "name": "OpenRouter", "verdict": "up", "status_url": "https://whatsdownnow.com/status/openrouter" }
]
}
Existing fields stay stable. payload_version is 2 when the new fields are present. Slack (hooks.slack.com) and Discord (discord.com/api/webhooks/...) URLs are auto-formatted; you can also set format to json, slack or discord when you add a webhook.
Verify the signature. Header X-WDN-Signature: t=<unix>,v1=<hex>, where v1 = HMAC-SHA256(signing_secret, t + "." + raw_body). The signing secret is returned when you create or read the stack. Reject timestamps older than 5 minutes.
import crypto from "node:crypto";
function verify(rawBody, header, secret) {
const { t, v1 } = Object.fromEntries(header.split(",").map((p) => p.split("=")));
if (Math.abs(Date.now() / 1000 - Number(t)) > 300) return false;
const mac = crypto.createHmac("sha256", secret).update(t + "." + rawBody).digest("hex");
return crypto.timingSafeEqual(Buffer.from(mac), Buffer.from(v1));
}
Respond with a 2xx status within 8 seconds. Failed deliveries are retried after about 5, 15, 30, 60 and 120 minutes, 6 attempts in all. A channel is disabled after 15 failures in a row or an HTTP 410.
MCP server for AI agents
Remote MCP endpoint over Streamable HTTP. It's stateless and needs no auth: https://whatsdownnow.com/mcp
Want your agent to set up alerts for you? Copy the one-line instruction.
get_status(slug) | One service's current status |
list_services(query?, category?) | Find a service by name, or discover slugs |
check_stack(slugs[]) | Readiness check: overall verdict and ready=true/false |
create_alert_stack(services, email?, webhook_url?, …) | Set up outage alerts; returns id, token and a private manage link |
get_alert_stack · update_alert_stack · send_test_alert · rotate_alert_stack_key · delete_alert_stack | Manage a stack with its id and token (rotate_alert_stack_key issues a new manage link; the old one stops working) |
list_incidents(slug) | Recent incidents and verdict changes |
request_service(name, url?, reason?) | Request we start monitoring a service that isn't in the catalog yet |
get_ai_fallback(provider?) | Healthy AI API alternatives, or the whole board if you omit provider |
report_issue(service, error_type, model?, region?) | Tell us about a 5xx, timeout or rate limit your agent just hit |
Copy-paste setup:
// Cursor: ~/.cursor/mcp.json · Claude Desktop: claude_desktop_config.json (or Settings › Connectors)
{ "mcpServers": { "whatsdownnow": { "url": "https://whatsdownnow.com/mcp" } } }
claude mcp add --transport http whatsdownnow https://whatsdownnow.com/mcp
// VS Code: .vscode/mcp.json
{ "servers": { "whatsdownnow": { "type": "http", "url": "https://whatsdownnow.com/mcp" } } }
curl -s https://whatsdownnow.com/mcp -H 'content-type: application/json' -H 'accept: application/json, text/event-stream' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"check_stack","arguments":{"slugs":["chatgpt","claude","stripe"]}}}'
Embed this badge
Add this badge to your README or site. https://whatsdownnow.com/badge/{slug}.svg shows a service's live status (up, degraded or down), refreshes about every 5 minutes and links back to its status page. Every service page has a ready-made snippet.
[](https://whatsdownnow.com/status/chatgpt)
<a href="https://whatsdownnow.com/status/chatgpt"><img src="https://whatsdownnow.com/badge/chatgpt.svg" alt="ChatGPT status on WhatsDownNow" height="20"></a>
Limits
- Up to 90 services, 5 webhooks, 3 email addresses and 10 browser subscriptions per stack.
- Creating stacks is limited to 5 per hour and 20 per day per network.
- Alerts are evaluated every 5 minutes, about 2 minutes after each status check. Each service is re-checked about every 11 minutes.
- Email alerts have a site-wide daily budget. If it runs out, email alerts pause until the next UTC day; webhooks and browser alerts keep working.
Service names and logos belong to their owners. WhatsDownNow is independent and not affiliated with any listed service.