[sentry-api-schema-updater]

No description provided.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
develop-docs	Ready	Preview	Oct 26, 2026 2:12pm UTC
sentry-docs	Ready	Preview	Oct 26, 2026 2:12pm UTC

[github-actions]

Please update the IS YOUR CHANGE URGENT? section of the PR description:

• Select exactly one option.
• If you selected an urgent or other deadline, provide the date as YYYY-MM-DD.

This information helps the Docs team prioritize your review.

[cursor]

The plan watches the projects field type on the create-release API docs page. It also watches API docs 404s, errors on API docs pages, and release API docs page loads. An issue escalates when those 404s or errors rise.

Services: sentry-docs.

Mention @change-monitor in a comment to update the plan.

Plan

What changed

The create-release API docs page now labels the projects body field as array(string). Today that page labels it array(undefined). The OpenAPI pin moved to 1aaa7af8 and types that field in two request-body copies. sentry-docs must deploy before the page updates. Shared src also builds into develop-docs. develop-docs does not render OpenAPI pages, so this plan does not watch it.

Risk

A bad schema can fail the production build. Then the site keeps the old pages. A slug change would 404 API docs pages. This diff only adds a string item type, so a build break is unlikely. If the API page component throws on the new type, array parameter labels on API endpoint pages can break. Readers of the create-release docs page would then see a blank page or a 404.

Intended effect

Confirmed means the create-release docs page labels projects as array(string). Absent means that page still labels projects as array(undefined), or the page returns 404. Page-load counts cannot prove the new label. Three loads of that endpoint in the last 24h only show the page still exists.

Signal	Baseline	Rule	Source
projects type on create-release docs page	array(undefined)	become array(string)	GET https://docs.sentry.io/api/releases/create-a-new-release-for-an-organization/ at 2026-10-26T14:07Z

Regression watch

This diff can 404 API docs pages or throw in the API page component. Watch 404 rate, API docs errors, and release API docs volume. At least one 404 or error rise on those pages is a regression. Ignore a single hour above 50 errors. That pattern matched a crawler burst at 2026-10-25T13:00Z.

Signal	Baseline	Rule	Source
API docs 404 rate	0.4-0.5% of API page loads. Counts were 2 then 4 per 24h. Denominator was 469 then 887 API page loads.	hold in 0.4-0.5%. Escalate above 1.0%.	Sentry metric docs.page.not_found query environment:production requested_path:api* over docs.page.load query environment:production page_type:api. Project docs. Windows 2026-10-24T14:07Z/2026-10-25T14:07Z and 2026-10-25T14:07Z/2026-10-26T14:07Z.
API docs error events	0-3 per 24h outside crawler hours. Current 24h count 2. Prior 24h count 4 plus a 295-event hour at 2026-10-25T13:00Z.	hold in 0-3 per 24h. Ignore one hour above 50.	Sentry errors query environment:production (http.url:/api/ OR transaction:/api/). Project docs. Same windows.
release API docs page loads	24-27 per 24h. Prior 27. Current 24.	hold in 15-40 per 24h. Escalate if the count falls below 10 while API 404s rise.	Sentry metric docs.page.load query environment:production page_type:api api_category:releases. Project docs. Same windows.
release API docs pageload failures	0 failures in 53 pageload and navigation spans	hold at 0	Sentry spans query environment:production span.op:[pageload,navigation] url.full:/api/releases/. Project docs. Window 2026-10-25T14:07Z/2026-10-26T14:07Z. failure_count()=0.

Not observable

Sentry metrics do not record the projects type string. Confirm that label with the live page GET in Intended effect. Prior-day failure count for release API docs pageloads was not returned as a series. Fleet-wide docs errors ran 860 in the current 24h and spiked above 3000 per hour from 2026-10-25T10:00Z to 2026-10-25T12:00Z. Those events are mostly unrelated promise rejections. Do not use fleet-wide error count. Pageload p95 on release API docs is sparse. Hours with traffic ranged 137-3675 ms. Pooled 24h p95 was 3180 ms. That spread is too wide for a tight latency rule.