Harden toolkit docs generation and indexing#1060
Conversation
Fail closed on unsafe or incomplete inputs, reuse one Engine snapshot, and expose generated routes to crawlers. Replace the redundant site build with focused generator validation and document the Vercel and Algolia flow. Co-authored-by: Cursor <cursoragent@cursor.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
@cursor review |
|
Current version of PR was reviewed by /review-bugbot with additional instructions on Jul 12, 22:44 GMT-3. It flagged 0 findings. Bugbot on commit |
Keep last-known-good output on failures while returning nonzero status, publish real toolkit routes to llms.txt, and align Markdown, sidebar, slug, and subpage handling across consumers. Co-authored-by: Cursor <cursoragent@cursor.com>
|
@cursor review |
Skip llms generation when its own bot commit synchronizes the PR, while retaining full-history diffs and fork safety. Co-authored-by: Cursor <cursoragent@cursor.com>
Ignore metadata-only changes so generated commits preserve the prior source SHA and do not trigger another artifact commit. Co-authored-by: Cursor <cursoragent@cursor.com>
|
@cursor review |
Allow manual workflow dispatches to target a dedicated branch and base ref without changing the shared automation PR. Co-authored-by: Cursor <cursoragent@cursor.com>
Request InstantSearch-compatible highlight markers and suppress stale results until they match the current typed query. Co-authored-by: Cursor <cursoragent@cursor.com>
Collapse rapid keystrokes into one request so the result list no longer repeatedly unmounts and rerenders during input. Co-authored-by: Cursor <cursoragent@cursor.com>
Exercise rapid query replacement and immediate clearing with fake timers so the anti-flicker request behavior cannot regress. Co-authored-by: Cursor <cursoragent@cursor.com>
Algolia search regression verificationReproduced on the preview before the fix:
Fixed by requesting Final generated preview measurement:
|
|
@cursor review |
Group generated integrations in llms.txt, reject invalid empty slugs, and preserve UI chunk placement and replacement semantics in Markdown exports. Co-authored-by: Cursor <cursoragent@cursor.com>
|
@cursor review |
Handle failed Algolia requests explicitly and preserve null safety when lightweight documentation chunks share sort priority. Co-authored-by: Cursor <cursoragent@cursor.com>
|
@cursor review |
Target the dispatch ref only for explicit verification branches while keeping shared automation PRs on the repository default branch. Co-authored-by: Cursor <cursoragent@cursor.com>
|
@cursor review |
There was a problem hiding this comment.
✅ Bugbot reviewed your changes and found no new issues!
Comment @cursor review or bugbot run to trigger another review on this PR
Reviewed by Cursor Bugbot for commit 63c423c. Configure here.
Split tracker
This implementation PR has been decomposed into reviewable source PRs. Do not merge this aggregate branch; review and merge the replacements in dependency order.
Independent PRs
Stacked PRs
Generated artifact
public/llms.txtis deliberately not copied into a source PR. Its diff in this aggregate branch contains pre-existing trailing whitespace; the idempotent generation workflow introduced in #1067 should create the artifact after the source behavior lands.Verification
On the original aggregate head:
pnpm run toolkit-docs:check— 46 files / 585 tests passedpnpm test— 60 files / 790 tests passedpnpm build— passedpnpm lint— exited 0 with three pre-existing complexity warnings in untouched generator filesEach replacement PR also has a focused test command recorded in its body. The workflow PR uses an isolated generator-test root so it remains independently green from
main.