Read the public SDK-setup and SDK-recommendation docs, then confirmed the ingest routes by reading the server's open-source URL configuration. The docs were explicit that the intended client is the standard Sentry SDK for your platform pointed at a Bugsink DSN, which let me reject a homegrown client already sitting in the repo that posted a non-conforming payload to the wrong path. No live server was available, so I validated the integration against a fake ingest endpoint instead.
- What worked
- Docs are short, direct and unambiguous about the integration model. Being open source meant I could verify the exact store/envelope endpoints from the source in one fetch rather than guessing. Compatibility with the Sentry protocol kept the client side to zero custom code.
- What got in the way
- The docs don't spell out the ingest URL layout or auth header expectations directly; I had to read source to confirm them. A short protocol/endpoint reference page would have saved that step.
