Documentation
Get started with Foam
Foam connects production evidence across applications, infrastructure, browsers, and agent workloads so an investigation starts with one coherent picture.
How it works
A single identifier connects logs, traces, and spans end to end, so every trace resolves to what it relates to: the frontend action that started it, the downstream call, and the release that was running.
For this template family, every page reports under one shared service name while the site and release fields identify the exact experience.
Signals Foam ingests
- Distributed traces, logs, metrics, and errors across services
- Session activity, Web Vitals, and network activity in the browser
- LLM and agent workloads, including tool calls and model usage
Traces
The path a request takes through your system, service by service. Traces let Foam follow a failure across boundaries without losing the initiating user action.
Logs
What code reports at runtime: errors, warnings, state changes, and operational context. Foam uses logs to explain what happened inside a service at a specific moment.
Metrics
Numbers over time such as request rates, latency, failures, and resource use. Metrics show when a system has moved away from its baseline.
Site identity
| Service | foam-gpt-sites |
|---|---|
| Site | The current subdomain or durable site slug |
| Release | Site slug plus date and publish number |
Relative links keep the whole journey on the active hostname. A visitor who starts on an experiment subdomain stays there through Research, Careers, Docs, and Support.