Help Guides
Operations (web)
The Operations section shows how the Platform API itself is running. It's in the web Hub only (the other apps skip these screens), it's read-only, and it's visible to superadmins. Once the API adds an operations.view permission (API issue #47), anyone holding it will see it.
Unlike the rest of the Hub, most of this data doesn't come from the Platform API. It comes from AWS (CloudWatch metrics and logs, alarms, budgets and cost) through the web Hub's own Ops gateway. The gateway checks your permissions with the API on every request and only ever reads.
The screens
| Screen | Shows |
|---|---|
| Overview | API health, alarms firing, requests, error rate, p95 latency, database capacity, webhook backlog, and month-to-date cost against budget |
| Traffic & latency | The same over time, against the API's service-level objectives: 99.9% of requests without a 5xx; p95 under 150 ms for token and permission checks, and under 400 ms for everything else |
| Logs | Saved log searches across the API's functions: by request ID, user, key, status or error code |
| Security events | Failed sign-ins, lockouts, refresh-token reuse, refused requests and blocked destructive calls |
| Alarms | Every alarm, its state and history, and who gets alarm emails |
| Jobs & delivery | The API's scheduled jobs, and webhook and push delivery health |
| Releases | Which version of each API function is deployed, with links to the release, deploy and rollback workflows |
| Cost | Month-to-date cost and forecast against the budgets |
| Configuration | The effective, non-secret configuration of each function, and where each value is set |
| Feature flags | The Hub's feature flags |
What's waiting on the API
- Per-operation traffic needs one log line per request (#43). Until then, Traffic & latency shows the API as a whole.
- Reports, with usage by key and by Application, needs usage rollups (#44).
- Editable flags and platform settings need #48. Until then, flags change with a deploy.
Integrations
Integrations collects the ways to connect other tools:
- platform-wide webhooks
- service and agent API keys
- the MCP server at
https://api.substratalapps.com/mcp, for AI assistants - where alarm notifications go
Sending logs to an external tool is configured in the API's infrastructure, not here.
Screens in this guide
OPS-01 · Overview OPS-02 · Traffic & latency OPS-03 · Logs SEC-01 · Security events OPS-04 · Alarms OPS-05 · Jobs & delivery OPS-06 · Releases OPS-07 · Cost CFG-01 · Configuration CFG-02 · Feature flags RPT-01 · Reports INT-01 · Integrations