agentleFS
diriendirien/yet-another-agent-harness/.claude/skills/dep-upgrade/SKILL.md

dep-upgrade

Skill19 starsChanged 8 months agoMITView on GitHub

Safe systematic dependency upgrade with vulnerability scanning and rollback capability. Detects project ecosystem, audits outdated and vulnerable packages, presents a prioritized upgrade plan, and executes upgrades one at a time with test verification after each. Use when updating dependencies, fixing vulnerability alerts, or performing periodic dependency maintenance.

4 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

4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If no package manager file is found, report to the user and stop.

    line 35
  2. Distinctive:

    If a vulnerability scanner is not installed, note it in the report but continue with the outdated package list.

    line 82
  3. Distinctive:

    After all upgrades are complete, verify lockfile integrity:

    line 146
  4. Distinctive:

    Do not commit the changes.

    line 176

“Common” is measured: kept by at least 6 in 10 of the 4 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- Copyright 2025-2026 Richard Shade. Licensed under Apache-2.0. --><!-- SPDX-License-Identifier: Apache-2.0 -->

Safe Dependency Upgrade

Systematically upgrade dependencies with zero breakage. Detects the project ecosystem, audits for outdated and vulnerable packages, presents a prioritized plan for approval, then executes upgrades one at a time with test verification after each.

Core principle: One upgrade at a time. Test after each. Roll back on failure. Never auto-commit.

Prerequisite check

Detect the project ecosystem by scanning for package manager files:

ls go.mod package.json pyproject.toml Cargo.toml *.csproj 2>/dev/null

If no package manager file is found, report to the user and stop. Mixed projects (e.g., Go backend + Node frontend) are supported — audit and upgrade each ecosystem separately.

Step 1: Audit dependencies

List outdated packages and run vulnerability scanners for each detected ecosystem.

Go:

go list -u -m all 2>/dev/null | grep '\['
govulncheck ./... 2>/dev/null

Node.js:

npm outdated 2>/dev/null
npm audit 2>/dev/null

Python:

pip list --outdated 2>/dev/null
pip-audit 2>/dev/null

Rust:

cargo outdated 2>/dev/null
cargo audit 2>/dev/null

.NET:

dotnet list package --outdated 2>/dev/null
dotnet list package --vulnerable 2>/dev/null

For detailed commands per ecosystem, see references/upgrade-commands.md.

If a vulnerability scanner is not installed, note it in the report but continue with the outdated package list.

Step 2: Categorize and prioritize

For each outdated package, assign a priority:

  • CRITICAL — known CVE or security advisory
  • HIGH — major version behind or dependency EOL
  • MEDIUM — minor version behind
  • LOW — patch version behind only

For detailed priority signals and breaking change assessment, see references/priority-matrix.md.

Step 3: Present upgrade plan

Present the prioritized list to the user before any upgrades. Use the plan format in references/report-template.md.

Dependency Upgrade Plan

Ecosystem: Go
Outdated: 5 packages
Vulnerable: 1 package

| # | Package           | Current | Target | Priority |
|---|-------------------|---------|--------|----------|
| 1 | example/vuln-pkg  | v1.2.0  | v1.2.5 | CRITICAL |
| 2 | example/old-pkg   | v2.0.0  | v4.1.0 | HIGH     |
| 3 | example/minor-pkg | v3.1.0  | v3.4.0 | MEDIUM   |

Proceed with upgrades?

Wait for user approval before proceeding. The user may choose to upgrade all, select specific packages, or skip.

Step 4: Upgrade one at a time

For each approved package, in priority order:

  1. Research — check changelog or release notes for breaking changes. If crossing a major version, read the migration guide.
  2. Upgrade — run the ecosystem-specific upgrade command
  3. Update lockfile — go mod tidy, npm install, etc.
  4. Run tests — execute the project's test suite
  5. If tests pass — move to the next package
  6. If tests fail — roll back this upgrade immediately:
# Rollback: restore dependency files from git
git checkout -- <dependency-file> <lockfile>

Report the failure (package, target version, test error) and continue with the remaining packages.

For ecosystem-specific upgrade and rollback commands, see references/upgrade-commands.md.

Step 5: Verify lockfiles

After all upgrades are complete, verify lockfile integrity:

# Go
go mod verify

# Node
npm ci --dry-run

# Rust
cargo build --dry-run

Confirm that lockfiles are updated and consistent with the dependency manifest.

Step 6: Present summary

Present the upgrade results using the template in references/report-template.md. Include:

  1. Successfully upgraded — package, from version, to version, test status
  2. Failed and rolled back — package, target version, failure reason
  3. Skipped — package, reason (user chose to skip, or breaking change needs manual migration)
  4. Remaining vulnerabilities — any CVEs not resolved
  5. Lockfile status — updated and verified per ecosystem

Do not commit the changes. The user decides when and how to commit. Suggest a commit message:

deps: upgrade X packages (Y security fixes)

Important constraints

  • Never auto-commit — upgrading packages changes files, but committing is the user's decision
  • Never create branches — leave git operations to the user
  • One at a time — upgrading multiple packages simultaneously makes it impossible to identify which upgrade broke tests
  • Roll back immediately on failure — do not continue with a broken dependency state
  • Report everything — even if the user chose to skip a package, include it in the summary with the reason

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.