agentleFS
Sign inStart free
jrittelmeyerjrittelmeyer/ai-dev-kit/.claude/skills/dep-check/SKILL.md

dep-check

Skill1 starsChanged 27 days agoMITView on GitHub

Registry-verify a dependency before adding or upgrading it — check the real published version, tags, and release age; enforce a release-age window; exact-pin frequent publishers; sanity-check peers and maintenance. Use when adding or bumping a dependency, picking a version, or triaging a Renovate/Dependabot PR. Never pick versions from blog posts or model memory.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use the adapter's ecosystem.registryQuery when set.

    line 19
  2. Very specific:

    Prefer the newest version older than the window (default ~7 days, adapter depPolicy.releaseAgeDays).

    line 30
  3. Distinctive:

    Graceful degradation: if the dep introduces an env-gated integration, the project must still build and run with that env unset — verify it.

    line 52
  4. Distinctive:

    Read the changelog/release notes for breaking changes on any major (and on fast-moving minors).

    line 58
  5. Distinctive:

    Check the project's memory/docs for standing holds: before merging a major — some majors are deliberately gated on an upstream event.

    line 60

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

dep-check

Model memory and blog posts are stale the day they're written; the package registry is the only source of truth for what a dependency's current version is. This skill front-loads a two-minute check that prevents the expensive failure modes: installing a just-published broken release, a prerelease masquerading as current, an abandoned package, or a silent breaking major.

Adapter: .claude/ai-dev-kit.config.json (depPolicy, ecosystem, legacy packageManager); defaults: 7-day window, judgment-based pinning.

1. Query the registry, not the web

  • Use the adapter's ecosystem.registryQuery when set; otherwise the per-ecosystem recipes in references/registries.md (npm-family · PyPI · crates.io · Go · NuGet · Maven · RubyGems · Composer · engine asset stores) — pick the section for the repo's manifest files.
  • Get three facts: the current stable version, the tag/channel landscape (latest can lag or lead; next/beta/canary/pre-releases must never be installed by accident), and the candidate's publish date.
  • One registry query beats any number of web searches for "what's the latest X".

2. Apply the release-age window

  • Prefer the newest version older than the window (default ~7 days, adapter depPolicy.releaseAgeDays). A just-published release can still be unpublished, yanked, broken, or — worst case — hijacked; a week of ecosystem soak is cheap insurance.
  • Exception: the newer release contains a fix you specifically need → take it and say so explicitly in the commit/PR body.

3. Pin policy

  • Exact-pin packages that publish frequently (adapter depPolicy.exactPin, plus judgment: multiple releases per month → pin exact) — a loose range on a fast publisher makes every fresh install a lottery.
  • Otherwise follow the project's prevailing range style (read a few neighbors in the dependency manifest).

4. Sanity checks before installing

  • Peers/host range: the package's declared peer or host-framework ranges vs what the project has installed.
  • Deprecation: the registry's deprecation/yank flag — a deprecated package needs a replacement, not an install.
  • Maintenance: last-publish date and repo activity as an abandonment proxy.
  • Graceful degradation: if the dep introduces an env-gated integration, the project must still build and run with that env unset — verify it.

5. Renovate / Dependabot PR triage

  • Re-check release age at merge time, not PR-open time.
  • Read the changelog/release notes for breaking changes on any major (and on fast-moving minors).
  • Check the project's memory/docs for standing holds before merging a major — some majors are deliberately gated on an upstream event; merging them early re-breaks a settled decision.

6. Record the decision

State in the commit/PR body: the version chosen, the pin style, the release date, and why (window respected / fix-needed exception / hold respected).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free