agentleFS
rjdelleceserjdellecese/confect/.agents/skills/upgrade-internal-deps/SKILL.md

upgrade-internal-deps

Skill382 starsChanged 6 days agoISCView on GitHub

Upgrade internal-only dependencies (toolchain devDependencies and the private workspace packages' deps) and open a PR—no changeset

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Every package.json dependency that is not claimed by the upgrade-published-deps skill (whose scope covers the published packages' …

    line 15
  2. Very specific:

    Verify the native type-benchmark harness too.

    line 41
  3. Distinctive:

    If nothing gets applied, say so and stop—no branch, no PR, even if some upgrades were spotted and deferred.

    line 47
  4. Very specific:

    Verify with the full repo checks (pnpm check, pnpm test, pnpm build).

    line 50

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Upgrade the dependencies that consumers of the @confect/* packages can never see, and open a PR for review. Never merge it yourself.

Scope

Exclude convex, convex-test, convex-helpers, and @convex-dev/* packages everywhere in the workspace, even when used only in private apps or fixtures. The audit-convex skill owns their upgrades and feature assessments.

Every package.json dependency that is not claimed by the upgrade-published-deps skill (whose scope covers the published packages' dependencies/peerDependencies plus their lockstep companions, such as react-dom with react): the workspace's devDependencies (build/test/lint toolchain, types) and the dependencies of every private workspace package—the apps under apps/* and the server-test fixture workspaces. Skip published-surface deps here even where they appear as devDependency pins, since their pins move with their ranges.

Rules

  • Discover updates with pnpm outdated -r and bump to latest. Don't restate or hand-enforce the workspace's pinning conventions—apply the bumps, then run pnpm lint:fix (Syncpack) to normalize.
  • Majors are fair game: attempt them, and if one requires more than mechanical changes to get green, drop it from the batch and explain what it would take in the PR description (or in your final report, if the run ends up applying nothing and opens no PR).
  • Upgrade @effect/tsgo, tsrs, and oxlint as a compatible set from the Effect release's supported-component table. Lint and Zed launch through the workspace's effect-tsgo proxy, so no separate editor version pin is needed. The prepare script patches only tsrs, for compiler scripts, Vitest, and the native benchmark harness. Every Effect rule except the intentionally disabled missing-pipeable-signature must stay enabled as an error, including newly introduced rules; scripts/effectLint.test.ts checks the installed schema against the configuration. Fix new findings and review any new export-specific unstable-API grants rather than disabling rules. Verify the native type-benchmark harness too; its patched dependency and baselines are documented in tools/bench-harness/README.md.
  • These upgrades are not user-facing: no changeset.

Delivering

  1. If nothing gets applied, say so and stop—no branch, no PR, even if some upgrades were spotted and deferred; the next scheduled run will surface them again.
  2. Verify with the full repo checks (pnpm check, pnpm test, pnpm build). Anything the local environment genuinely can't run, leave to the PR's CI—and get it green.
  3. Push a branch (deps/<short-description>, unless this session was assigned a branch) and open a PR against main. In the body, list what was bumped and note anything deliberately skipped and why.

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.