fix(knip): scope packages/* entry exports so dead code gets flagged

[bbornino]

Closes #204

Problem

knip's default entry-file exemption treats every package's declared main (src/index.ts) as a public API boundary — reasonable for a package with real external consumers, but wrong for packages/* here, since these are all private: true internal workspace packages with no consumer outside this monorepo. As a result, knip wasn't catching genuinely dead exports in these packages, which is how a dead list() function and S3Object interface in packages/s3/src/utils.ts went unflagged through multiple clean test:unit runs.

Fix

One-line addition to knip.json, matching the existing apps/worker override pattern:

"packages/*": {
      "includeEntryExports": true
}

Cleanup

Running knip with the fix surfaced 12 previously-invisible findings. Investigated each individually (checked for real consumers anywhere, including tests, and whether any might be reserved for planned-but-not-yet-wired-up work like #94's schema endpoints) rather than blanket-ignoring the list:

Un-exported only (used internally, so kept — just no longer part of the package's public surface):
- DELETE_S3_OBJECT_GRACE_PERIOD_MS, PostImageInputSchema, TasksKeys (packages/bullmq)
- EnvSchema (packages/common)
- FIRST_CONTRIBUTOR_YEAR (packages/github-api)

Removed entirely (zero consumers anywhere):
- TaskInputsValues / TaskOutputsValues (packages/bullmq) — never referenced anywhere, including internally; only TaskInputs/TaskOutputs themselves are actually used
- EnvType (packages/common) — never referenced
- getCommits / GetCommitsParams (packages/github-api) — whole file deleted. This REST-based commit listing was never wired up to any caller; getAuthorGitHubStats.ts already covers commit-history stats via a separate GraphQL query
- exists / matchesEtag (packages/s3) — exists had no callers; matchesEtag's ETag-based staleness check was superseded by the LastModified/unmodifiedSince approach from #197's delete-s3-object work (ETag can't detect a same-content rewrite, which is exactly the race that check needs to catch)

One edge case worth flagging for review: un-exporting PostImageInputSchema revealed it's a const whose only usage anywhere is Static<typeof PostImageInputSchema> — a type-only query, never read as a runtime value. ESLint's no-unused-vars correctly flags this once it's no longer exported (exports are exempt from that check). Resolved with a scoped eslint-disable-next-line and a comment explaining why, rather than restructuring or leaving it exported (which would just re-trigger the knip finding this PR fixes).

Testing

- pnpm exec knip — 0 findings (was 12 with the config change alone, 0 before the config change)
- pnpm build:all — clean, all 10 projects
- pnpm test:unit — clean (sherif, eslint, publint, vitest all green across every project)

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: eca206b5-30d3-4868-bd8c-ee525d64bd44

---

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.}