document-api-endpoint — Claude Code skill: what it does and setup
Documents and types Sentry API routes using drf-spectacular. It resolves type drift between declared OpenAPI schemas and live endpoint responses.
npx skills add getsentry/skills --skill document-api-endpoint -a claude-codeWhat document-api-endpoint does
The skill guides adding @extend_schema decorators to Sentry endpoint classes and methods. It reuses parameters from src/sentry/apidocs/parameters.py, assigns tags from OPENAPI_TAGS, and builds response shapes with TypedDict. For payload fields, T | None signals a required key with a null value, while NotRequired[T] handles optional keys like those triggered by an expand query parameter.
Migrating legacy paths is strictly all-or-nothing per path because drf-spectacular's APPEND_PATHS does not merge HTTP methods. Adding @extend_schema to one method causes all untransferred legacy methods on that route to disappear from the schema. Moving an endpoint to PUBLIC requires updating publish_status, assigning owner = ApiOwner.<TEAM>, and clearing the method from API_OWNERSHIP_ALLOWLIST_DONT_MODIFY.
Verification relies on make build-api-docs, pnpm run validate-api-examples, and targeted pytest runs in the virtual environment. To catch discrepancies like numbers formatted as strings, compare live curl output piped into jq against your types.
What it helps with
- add
@extend_schemadecorators to a Sentry endpoint - fix type drift between runtime API responses and schemas
- migrate legacy JSON specifications from
api-docs/paths/ - promote a private endpoint to
PUBLICwith team ownership
Example prompts for the agent
- Document this Sentry endpoint and write the response TypedDict
- Fix type drift for this route against the live API response
- Promote this endpoint to PUBLIC and remove it from the ownership allowlist
How to install document-api-endpoint
1Skills CLI (Vercel)
npx skills add getsentry/skills --skill document-api-endpoint -a claude-code2Manually: put the files in place
~/.claude/skills/document-api-endpoint/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/document-api-endpoint/
1Skills CLI (Vercel)
npx skills add getsentry/skills --skill document-api-endpoint -a codex2Manually: put the files in place
.agents/skills/document-api-endpoint/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add getsentry/skills --skill document-api-endpoint -a cursor2Manually: put the files in place
.cursor/skills/document-api-endpoint/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install getsentry/skills document-api-endpoint2Manually: put the files in place
.github/skills/document-api-endpoint/SKILL.md1Skills CLI (Vercel)
npx skills add getsentry/skills --skill document-api-endpoint -a gemini-cli2Platform CLI
gemini skills install https://github.com/getsentry/skills --path skills/document-api-endpointManually: put the files in place
.devin/skills/document-api-endpoint/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add getsentry/skills --skill document-api-endpoint -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about document-api-endpoint
How to install document-api-endpoint in Claude Code?
Run npx skills add getsentry/skills --skill document-api-endpoint -a claude-code in your terminal.
What does document-api-endpoint do?
It provides instructions for documenting Sentry endpoints with drf-spectacular. It covers response typing, decorator setup, parameter reuse, and schema validation.
Why must legacy api-docs paths be migrated all at once?
drf-spectacular's APPEND_PATHS does not merge HTTP methods on the same route. Once any method uses @extend_schema, all legacy methods on that path vanish unless migrated in the same commit.
Source text
SKILL.md · Apache-2.0
---
name: document-api-endpoint
description: Document and type a Sentry API endpoint. Write or fix @extend_schema decorators, specify response TypedDicts, type request parameters, correct type drift between the declared schema and the runtime response, and validate the generated spec. Use when asked to "document an endpoint", "add OpenAPI docs", "add/fix @extend_schema", "type an endpoint response", "fix the response type", "fix type drift", "reuse a response type", "split an overloaded endpoint", "specify the response schema", "add a TypedDict response", "migrate a legacy api-docs path", "fix a parameter type", or "make an endpoint public" / "promote an endpoint" (promotion is one section here).
---
# 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.
```bash
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
```bash
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>
```
© getsentry · Apache-2.0 · getsentry/skills