Concepts

API keys

Every batch sent to API Glimpse needs a valid project API key (ask_…).

• Create keys in the dashboard (shown once when you create them)
• Set API_SENSOR_KEY in your app
• Missing or invalid key → batch rejected; endpoints are not updated
• The key maps to a project; endpoint data stays in that project

Schemas and tags

API Glimpse stores schemas, counters, and tags — not long-lived raw request or response bodies.

• Connectors remove secrets (authorization, cookies, and similar) before data leaves your app
• Body field names and types may be inferred from samples
• Samples exist briefly for aggregation, then are discarded

Endpoint limits

A project may have an endpoint limit:

• New endpoints over the cap are not added
• Existing endpoints continue to update as traffic arrives

Path templating

Stable path templates via classifiers (UUID, numeric, hex, email, opaque tokens) plus vocabulary heuristics so /users/1 and /users/2 collapse without turning static paths into noise.

Schema merge

Required fields = intersection across observations; types widen to unions; property fan-out is capped. Merged fragments become the latest view of each endpoint’s fields.

Sampling and availability

Sampling is asynchronous: the connector does not wait on API Glimpse before finishing your request. If API Glimpse is unreachable, samples drop and your app continues to serve traffic. See Connect your app → Troubleshooting.

Protect mode (later)

Runtime request blocking is planned for a later release and is not enabled today. API Glimpse v1 is discovery: endpoints, schemas, and tags.