agentleFS
GavinGreenwoodGavinGreenwood/agentic-workflow-template/.agents/skills/dependabot-review/SKILL.md

dependabot-review

Skill5 starsChanged 45 days agoMITView on GitHub

User-invoked only. Merge green minor/patch bumps, diagnose failing ones, escalate majors

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

10 rules. About 0 are common, 10 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    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 there are no open Dependabot PRs, output: "✅ No open Dependabot PRs." and stop.

    line 19
  2. Distinctive:

    Check if it is a major version bump (breaking change risk):

    line 39
  3. Distinctive:

    If major → print a warning and ask: "PR #<number> is a major bump (<package> <old> → <new>).

    line 40
  4. Distinctive:

    Wait for confirmation before proceeding.

    line 41
  5. Distinctive:

    If minor or patch → proceed automatically.

    line 42
  6. Distinctive:

    Ask: "Should I check out PR #<number> and attempt to fix it, or close it and let Dependabot re-raise later?"

    line 79
  7. Distinctive:

    If fix → check out the branch, apply the fix, push, and wait for CI to re-run.

    line 80
  8. Distinctive:

    If close → close the PR with an explanation comment and let Dependabot re-raise when the ecosystem catches up:

    line 81
  9. Distinctive:

    Default to close when: it is a major bump with ecosystem-wide incompatibility, the fix requires significant code changes, or the upstream …

    line 85
  10. Distinctive:

    Default to fix when: it is a minor/patch bump, the fix is a one-liner (e.g.

    line 86

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Review and action open Dependabot PRs

Fetch all open Dependabot PRs for this repo, check their build status, and approve + merge the green ones. Diagnose and action the failing ones.


Step 1 — List open Dependabot PRs

gh pr list --author "app/dependabot" --state open --json number,title,url,headRefName,createdAt,labels --limit 50

If there are no open Dependabot PRs, output: "✅ No open Dependabot PRs." and stop.

Step 2 — For each PR, fetch build status

gh pr checks <number> --json name,state,conclusion,link

Classify each PR as:

  • Green — all checks passed
  • Failing — one or more checks failed
  • Pending — checks still running
  • No checks — no CI configured for this PR (treat as Failing — do not merge)

Step 3 — Action Green PRs

For each Green PR:

  1. Review the PR title to understand what is being updated (package name, version bump, major/minor/patch).
  2. Check if it is a major version bump (breaking change risk):
    • If major → print a warning and ask: "PR #<number> is a major bump (<package> <old> → <new>). Approve and merge anyway?"
    • Wait for confirmation before proceeding.
    • If minor or patch → proceed automatically.
  3. Approve the PR:
    gh pr review <number> --approve --body "Dependabot auto-approval: all checks green, patch/minor bump."
  4. Merge the PR using squash merge:
    gh pr merge <number> --squash --delete-branch
  5. Confirm merge succeeded.

Step 4 — Action Failing PRs

For each Failing PR:

  1. Identify which checks failed:
    gh pr checks <number>
  2. Fetch the failed job logs via the gh CLI (see fix-cicd in .agents/skills/fix-cicd/SKILL.md for the drill-down pattern — use the run ID from the checks output). Read the logs and identify the root cause.
  3. Categorise the failure:
    • Dependency conflict — the new version conflicts with another package
    • Type error — the new version changed a type signature
    • Test failure — a test broke due to changed behaviour
    • Lint failure — new version introduced a linting issue
    • Flake / transient — looks like a non-deterministic failure
    • Unknown
  4. For Flake failures → re-trigger the checks:

    gh pr comment <number> --body "@dependabot rebase"
  5. For fixable failures (conflict, type error, test, lint):
    • Describe the fix needed (1–3 sentences).
    • Assess whether the fix is straightforward (one-liner, clear root cause) or complex (needs research, risky).
    • Ask: "Should I check out PR #<number> and attempt to fix it, or close it and let Dependabot re-raise later?"
    • If fix → check out the branch, apply the fix, push, and wait for CI to re-run.
    • If close → close the PR with an explanation comment and let Dependabot re-raise when the ecosystem catches up:
      gh pr close <number> --comment "## Dependabot Review\n\n**Closing** — <reason the fix isn't viable now, e.g. upstream incompatibility>. Dependabot will re-raise this once <condition>."
    • Default to close when: it is a major bump with ecosystem-wide incompatibility, the fix requires significant code changes, or the upstream package hasn't yet released a compatible version.
    • Default to fix when: it is a minor/patch bump, the fix is a one-liner (e.g. adding a missing param), and the change is low-risk.
  6. For Unknown failures → leave a diagnostic comment and flag to the user.

Step 5 — Pending PRs

For PRs where checks are still running, note them and skip. They will be picked up next time this command runs.

Step 6 — Final report

Print a summary table:

| PR | Package | Bump | Status | Action taken |
|----|---------|------|--------|--------------|

And a one-line summary: "Merged X, diagnosed Y, skipped Z (pending)."

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.