CoffeeRelay Documentation
Webhook relay built for AI agents and event-driven workflows
For AI Agents — Setup Guide
Current Status: API keys are not yet available. Endpoint and destination management currently requires dashboard access via GitHub OAuth.
Coming in Phase 1: Bearer token auth, OpenAPI spec, TypeScript SDK, and official MCP server.
What Works for Agents Today
✅ Public Ingest
POST to /api/in/{slug} — no auth required
✅ Signature Verification
Every forwarded webhook includes x-coffeerelay-signature
✅ Session State
Webhooks with x-session-id get state hydrated
✅ Idempotency
Send idempotency-key header for deduplication
Setup Steps (Current Manual Process)
- Human creates account and endpoint
- Sign up at /sign-in (GitHub OAuth)
- Dashboard → Endpoints → New Endpoint
- Copy slug and signing secret
- Human adds destination(s)
- Endpoint detail page → Destinations tab
- Add destination URL where agent receives webhooks
- Copy ingest URL:
https://coffeerelay.dev/api/in/{slug}
- Register ingest URL with external service
- Paste ingest URL into Stripe/GitHub/etc webhook settings
- External service will POST events to CoffeeRelay
- CoffeeRelay forwards to your destination(s) in real-time
- Agent verifies signatures
- Check
x-coffeerelay-signature header on received webhooks - Use signing secret from dashboard to verify HMAC-SHA256
- See llms-full.txt for code examples
Test Your Setup
curl -X POST https://coffeerelay.dev/api/in/your-slug \
-H "Content-Type: application/json" \
-d '{"event": "test", "data": "hello"}'
Expected response:
{
"received": true,
"logId": "clx...",
"destinations": 2
}
Core Features
Multi-Destination Fan-Out
One webhook in → multiple destinations receive it instantly in parallel
Session State Storage
Persist agent context across webhook events for multi-step workflows
Circuit Breaker
Auto-pause on error spikes to prevent expensive agent retry loops
Dead Letter Queue
Capture failed deliveries for bulk replay after incidents (Pro)
Idempotency Keys
Deduplicate webhooks to prevent expensive duplicate LLM calls
Policy Rules
Claude Code safety with allow/deny patterns for agent tool use
Plans & Pricing
Free
$0
- 1 endpoint
- 1 destination
- 1K requests/month
- 1-day log retention
Starter
$10/mo
- 5 endpoints
- 3 destinations each
- 10K requests/month
- 7-day logs
- Replay + circuit breaker
- 50 sessions/endpoint
Pro
$25/mo
- 25 endpoints
- 10 destinations each
- 100K requests/month
- 30-day logs
- DLQ + unlimited rules
- 500 sessions/endpoint
Why Event-Driven for AI Agents?
- Saves tokens and compute — Agents only activate when real events arrive, no empty polling requests
- Near real-time reactions — Respond instantly to payments, code pushes, support tickets
- Lower API costs — 1 request per event vs N polling requests
- Simpler architecture — No cron jobs, timers, or polling loops to manage
Need Help?
Email: support@coffeerelay.dev
Complete Docs: llms-full.txt