perf(api): cache the live link item lookup for public item pages

[Zach Dunn (zachdunn)]

In plain terms

A public live link item page (/c/<id>/<item>) used to find its item by listing the live link's whole scope and hashing every object key until one matched the item id. That ran on every request, with no sign-in and no cache. PR comments now link every file to its item page, so this path gets more traffic. This change keeps each live link's hashed scope list in a short-lived server-side cache, so most item page requests skip the scan.

What it does / what it is not

• Caches each live link's hashed scope list (item id, object key, timestamp) in a named Workers Cache API cache for 60 seconds. The cache key is the live link id plus its scope, so no list is shared across live links. The cache is server-side only and is never served to clients.
• The cached list is a hint, not the authority:
  • The route still resolves the live link in D1 first, so a revoked link 404s even while its list is cached.
  • An id found in the cached list is served only after a one-row D1 check that its key is still in scope. A deleted or re-scoped file fails the check, and the list is rebuilt.
  • An id not in the cached list runs one query for scope rows stamped at or after the list's newest row. The list is rebuilt only when one of those rows matches the id. A random or stale id gets a 404 without a full scan.
  • Hydration is unchanged, so private objects are still withheld, and missing objects are still missing.
• With no Cache API (unit tests, workers.dev previews), every lookup scans as before.
• Not changed: the public cursor shape {v,u,h}, the private count cap and its null semantics, title privacy, and hydrateFeedItems.
• Not done: the public cursor decode for GET /public/feeds/:id?cursor= already narrows its lookup to rows that share one exact timestamp (an indexed equality), so it does not do the 2,000-row scan. I left it as is.
• Not done: per-IP rate limiting. The web worker calls this endpoint over the API service binding with no client IP, so a per-IP key would put all SSR traffic in one bucket. A rate limit belongs at the edge on /c/* if one is needed.
• Known gap: a file that joins the scope without a new gh.repo timestamp (for example, a path row added later to an existing file) can 404 on its item page until the cached list expires (60 seconds at most).

Technical notes

• New module apps/api/src/live-link-index.ts (findLiveLinkItem). publicFeedItemPage now takes prev/next ids from the list instead of hashing them again.
• pr-scope.ts: scanScopeKeys takes since (updated_at >= ?, served by file_metadata_gh_repo_recent_idx), and the new scopeHasKey checks a single key.
• A stored entry carries its build time, and reads reject anything older than the TTL, even when the Cache API would still return it.
• No migration, no new binding, and no changeset (@uploads/api is ignored by changesets).

Test plan

• pnpm --filter @uploads/api test: new cases in routes-feeds.test.ts with an in-memory caches fake. They count scope queries to cover: a cached hit with no rescan, a 404 for an unknown id with one bounded query, a file uploaded after caching, a deleted file never served from the cache, a file made private after caching (withheld), list isolation between two live links, a revoked link, and TTL expiry.
• I confirmed that 5 of the new cases fail with the cache disabled. The isolation and TTL cases pass either way by design.
• Existing pager and cursor tests pass unchanged on the uncached path.
• pnpm --filter @uploads/api typecheck, pnpm check
• After deploy: open a few /c/<id>/<item> pages from a PR comment and confirm the neighbours and the withheld items look right.

Closes #1072

[changeset-bot]

⚠️ No Changeset found

Latest commit: 13e3357

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[coderabbitai]

Important

Review skipped

Auto reviews are limited based on label configuration.

🏷️ Required labels (at least one) (2)

• coderabbit:review
• review

🚫 Excluded labels (none allowed) (1)

• wip

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 37486e0d-d563-4930-bcab-9cd9b4720183

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	uploads-api	13e3357	Commit Preview URL Branch Preview URL	Oct 05 2026, 11:45 AM

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	uploads-web	13e3357	Commit Preview URL Branch Preview URL	Oct 05 2026, 11:45 AM