wacli.me
alpha
This page is maintained by hand, and it says so. An automated status page that always shows green is worth less than a sentence written by someone who knows what broke.
| Component | State | Notes |
|---|---|---|
MCP endpoint /mcp | running | Stateless, one process |
| Website | running | Static, served by the same process |
Health check /healthz | running | Version and tenant count |
| Sync daemons | manual | Started per account by hand; supervision is on the roadmap |
| Self-service linking | running | QR flow at /connect, token or OAuth |
| OAuth 2.1 | running | Dynamic registration, PKCE; no token to copy |
| Redundancy | none | Single host, single region |
| Automated monitoring | none | Failures are noticed, not alerted |
One machine runs everything. If it reboots, the endpoint is gone until the service comes back up. If the tunnel in front of it reconnects, requests fail for a few seconds. If a WhatsApp session expires, that one account returns empty results until it is re-linked — the endpoint itself stays healthy, which is why account_status exists.
There is no SLA, no uptime number, and no compensation. If that is not acceptable for what you want to do, run it yourself — then the machine is yours and so is the uptime.
no-cache.Incidents are also listed in the changelog, with the same wording. Nothing gets quietly dropped from one and kept in the other.