dep-upgrade
Skill4 starsChanged 2 days agoMITView on GitHub
Upgrade dependencies safely, one at a time: list what is outdated, read each changelog for breaking changes, bump one package (or one tight group), run the build and tests, fix or roll back, and commit before the next. Use for "update dependencies", "bump packages", "upgrade X to v5", "fix the outdated/vulnerable packages", "Dependabot PRs pile", or a framework major-version upgrade.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- dirien/yet-another-agent-harnesstheir own version, first seen Nov 8, 2026
- nuncaeslupus/literary-critictheir own version, first seen Oct 13, 2026
- AlpayC/claude-skill-settheir own version, first seen Sep 14, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- dirien/yet-another-agent-harness★ 19same name
- AlpayC/claude-skill-set★ 0same name
- nuncaeslupus/literary-critic★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository50 other files it gives its agents.
- agent-writingSkill
- arenaSkill
- blast-radiusSkill
- bun-playwright-runner-selectionSkill
- cinematic-shotSkill
- clean-ui-shotSkill
- codebase-designSkill
- debug-loopSkill
- diagnosing-bugsSkill
- discoverSkill
- domain-modelingSkill
- earshotSkill
- eval-judgeSkill
- flaky-testSkill
- grillingSkill
- grillSkill
- idea-scoutSkill
- improve-codebase-architectureSkill
- interrogateSkill
- make-mediaSkill
- migrate-to-rustSkill
- migration-planSkill
- multi-agent-windows-hazardsSkill
- perf-loopSkill
- postmortemSkill
- prototypeSkill
- prSkill
- real-soundSkill
- recallSkill
- release-notesSkill
- repo-mapSkill
- researchSkill
- retroSkill
- secure-diffSkill
- sellSkill
- subagent-report-recoverySkill
- tasteSkill
- taste-skillSkill
- tddSkill
- teachSkill
- to-questionnaireSkill
- to-specSkill
- to-ticketsSkill
- wayfinderSkill
- weeklySkill
- windows-bash-backslash-halvingSkill
- wizardSkill
- writing-for-agentsSkill
- ycSkill
- ystack-upgradeSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 30
Use the ecosystem's own tool (bun outdated, pnpm outdated, pip list --outdated, cargo outdated, go list -m -u all, npm audit / pip-audit …
- Distinctive: line 53
Run the same checks as the baseline.
- Distinctive: line 75
Never delete the lockfile to "fix" an install.
- Distinctive: line 76
Don't add new packages during an upgrade run.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Dependency upgrade
Bumping everything at once turns ten small breakages into one knot nobody can untangle. Upgrade one thing, prove it, keep it, move on.
When it applies
- Several outdated or vulnerable packages to bring up to date.
- A single major-version upgrade (a framework, a compiler, a runtime).
- Merging a backlog of automated bump PRs.
Procedure
1. Baseline
On a clean branch, run install, build, type check, lint and tests. Record what passes now. A test that already fails is not the upgrade's fault; note it and keep it out of the verdicts.
2. List and order
Use the ecosystem's own tool (bun outdated, pnpm outdated, pip list --outdated,
cargo outdated, go list -m -u all, npm audit / pip-audit for vulnerabilities). Order:
- security fixes;
- patch and minor bumps, batched only if they're in the same family;
- majors, one per step, low-level ones (runtime, compiler, test runner) before the things on top.
Packages that must move together (a framework and its plugins, a type package and its library) are one step.
3. Read before you bump
For each major, read the changelog or migration guide between the current and target version. Note every breaking change, then search the code for each removed or renamed API. If the list is long, split the upgrade into a prep step (adapt the code on the old version where possible) and the bump.
4. Bump one step
Change the one version, install, and let the lockfile update. Check the lockfile diff for surprise changes to other packages.
5. Prove it
Run the same checks as the baseline. Then exercise the code that uses the package: its tests, or a short real run of the feature. "It compiles" is not enough for a library with runtime behaviour.
6. Keep or roll back
- Green: commit with the package, old -> new version, and anything you adapted. One commit per step.
- Red and the fix is small and clear: fix, re-run, commit.
- Red after two fix attempts: roll back that step, note why, and move to the next package.
7. Report
| Package | From -> To | Result | Notes |
| lib-a | 2.1 -> 2.4 | kept | - |
| lib-b | 4 -> 5 | kept | adapted 3 call sites to new API |
| lib-c | 1 -> 2 | held | breaks X; needs <decision or work> |
Baseline failures (not caused by upgrades): <list>Rules
- One step per commit, so any later breakage bisects to one bump.
- Never delete the lockfile to "fix" an install. Resolve the conflict.
- Don't add new packages during an upgrade run.
- Pin and note a package you hold back; don't leave a range that will drift.
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
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.

