getsentry/document-api-endpoint

Document & Type a Sentry API Endpoint

Add or fix OpenAPI docs for a Sentry endpoint with drf-spectacular. Full reference is at https://develop.sentry.dev/backend/api/public/, the most useful section to you will be https://develop.sentry.dev/backend/api/public/#5-method-decorator. This skill captures the non-obvious lessons on top of it. Most of the work is making the declared schema match what the endpoint actually returns. Before documenting, identify which endpoint class serves the route and what it does; the MCP tool that calls it is usually the fastest way to confirm its behavior. Promoting a PRIVATE/EXPERIMENTAL endpoint to PUBLIC is one application (see below).

Workflow

1. Class-level @extend_schema(tags=[...]) — use the closest existing OPENAPI_TAGS entry.
2. Method-level @extend_schema(operation_id=..., parameters=[...], responses={...}, examples=...).
3. Reuse src/sentry/apidocs/parameters.py and examples/*.py; ensure owner = ApiOwner.<TEAM> is set.
4. If a legacy api-docs/paths/**/*.json covers the path, remove it (see lesson 4).
5. Validate, then verify against the live endpoint (lesson 1).

Lessons

1. Carefully compare what the code does vs declared types

Ideally, hit the live endpoint with a real token and diff the keys and types against your TypedDict. Serializers are sometimes inaccurate. Look out for counts coming back as floats instead of integers, IDs declared int emitted as strings, nested types declaring the wrong number of fields. Correct the declared type to match runtime.

curl -s -H "Authorization: Bearer $TOKEN" "https://us.sentry.io/api/0/<endpoint>" | jq 'keys'

2. Reuse the canonical response type

Match the codebase's XxxResponseOptional(TypedDict, total=False) mixin (main class declares required fields). Nullable-vs-absent: T | None = key always present, value may be null; NotRequired[T] = key only set under a condition (e.g. an expand query param). Reuse the existing canonical type instead of re-declaring a second or third copy in a *_types.py. If there's no clean canonical type to reuse (e.g. a payload proxied from another service like vroom/profiling), type it dict[str, Any] rather than inventing a new mirror, and confirm the shape from the owning service's repo, not just the serializer.

3. Infer the type. Avoid cast and # type: ignore

When a serializer returns a base type plus extra fields, refactor the producing code so the response type is inferred rather than forced.

4. Legacy doc migration is all-or-nothing per path

Delete the api-docs/paths/**/*.json file AND its $ref in api-docs/openapi.json. drf-spectacular's APPEND_PATHS does not merge HTTP methods, so once any method on a path uses @extend_schema, all legacy methods on that path vanish — migrate every method on the path in one commit.

Promoting to PUBLIC

Do the workflow above, then on the concrete endpoint only (leave siblings PRIVATE):

• Bump publish_status[<METHOD>] → PUBLIC and set owner = ApiOwner.<TEAM>.
• Remove the method from API_OWNERSHIP_ALLOWLIST_DONT_MODIFY in the same change as the flip.
• If the endpoint is redundant or being renamed, delete or deprecate the old version in its own change first, then stack the publish on top.
• Note in the PR if scopes widen (event:read → event:{admin,read,write}) — that's drf-spectacular regenerating from permission_classes, documentation-only.

The change reaches the @sentry/api SDK / MCP only after sentry-api-schema regenerates downstream.

Validate

make build-api-docs
pnpm run validate-api-examples
.venv/bin/pytest -q --reuse-db tests/apidocs/endpoints/<area>/test_<name>.py
.venv/bin/prek run -q --files <changed paths>