AntFleet

Disagreement · 3376e50a-openai-0

Unpublished dependency versions (TypeScript 6.0.3, Vitest 4.1.5) will break installs/builds

mismatch
repo 53606958·PR #4·reviewed 1 week ago

Primary finding

Unpublished dependency versions (TypeScript 6.0.3, Vitest 4.1.5) will break installs/builds

highbuild-releasehigh
  • package.json:23
  • package.json:24
As of the knowledge cutoff (2024-10), TypeScript major version 6 and Vitest major version 4 are not published on npm. Using non-existent versions causes package managers to fail resolution, preventing installation, builds, and tests from running.

Recommendation

Pin to known published versions. For example: typescript ^5.6.x (or current stable), vitest ^1.x or ^2.x (current stable). Verify with `npm view typescript versions --json` and `npm view vitest versions --json`, update package.json accordingly, and regenerate the lockfile.

Counterpart finding

Husky prepare script will fail on `npm ci --omit=dev` and fresh clones without husky installed

mediumbuild-releasemedium
  • package.json:15
  • package.json:19
`prepare` runs automatically after `npm install`. In production-style installs that skip dev dependencies (e.g. `npm ci --omit=dev` or when this template is consumed as a dependency), the `husky` binary will not be present and the lifecycle script will exit non-zero, breaking installs. For a template repo intended to be forked and deployed per-agent by an automated launchpad, this is a realistic failure mode. The conventional guard is `husky || true` or `husky install` gated on a CI env var.

Recommendation

Guard the prepare script, e.g. `"prepare": "husky || true"` or only run husky when not in CI/production: `"prepare": "node -e \"if(!process.env.CI)require('husky').default()\""` (or use husky's documented pattern).

Why this didn't post

This finding didn't meet AntFleet's unanimous agreement threshold. Both frontier models review every PR independently; only findings they both flag with the same severity and category are posted to the PR. This one fell through.

read the methodology →

From the same review

These findings passed the unanimous gate on the same PR review. The disagreement above was filtered out; the findings below were posted.