From sst
This skill should be used when the user mentions "dependabot", "dependency updates", "review dependency PRs", "check dependabot PRs", or asks about open PRs from Dependabot. Reviews open Dependabot pull requests for compatibility, urgency, breaking changes, and CI/CD status, producing a structured recommendation report. Default reviews all open Dependabot PRs; accepts specific PR numbers as arguments.
How this skill is triggered — by the user, by Claude, or both
Slash command
/sst:review-dependabot-prsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Analyze Dependabot PRs and produce an evidence-based recommendation report. Do NOT merge or close PRs — only report.
Analyze Dependabot PRs and produce an evidence-based recommendation report. Do NOT merge or close PRs — only report.
1755 1754): review only those PRs# All open dependabot PRs (default)
gh pr list --author "app/dependabot" --state open --json number,title,headBranch,labels,createdAt,url
# Or fetch specific PRs
gh pr view <number> --json number,title,headRefName,labels,createdAt,url,body,statusCheckRollup
If no open Dependabot PRs exist, report that and stop.
Launch one Agent subagent per PR (all in parallel). Each subagent receives the full context below and investigates a single PR.
For each PR, the subagent must gather and return ALL of the following:
A. PR & diff details
gh pr view <number> --json body,files,commits,statusCheckRollupgh pr diff <number> — read the actual diff to see what changedB. Lockfile check
pyproject.toml, package.json, Cargo.toml, etc.), check whether the corresponding lockfile was also updated in the same PRrequirements.txt, uv.lock, poetry.lock, package-lock.json, yarn.lock, pnpm-lock.yaml, Cargo.lock, etc.)C. CI/CD status (critical — do not skip)
statusCheckRollup, report each check's name, status, and conclusiongh run view <run-id> --log-failed and summarizeD. Dependency research (use Context7 MCP and/or WebSearch for current info)
E. Cross-dependency compatibility
F. Codebase impact scan
G. Confidence and evidence
After all subagents return, analyze interactions between PRs:
Produce a single structured report. Use this format for each PR:
## PR #<number>: <title>
**Package**: <name> | **From**: <old> | **To**: <new> | **Ecosystem**: <dir>
### Lockfile sync
⚠️ **Lockfile not updated** — manifest changed but lockfile was not updated in this PR
(or: ✅ Lockfile updated in PR / N/A — no manifest change)
### CI/CD Status
<status of each check — pass/fail/running>
### What changed (old → new)
- <key changes grouped: security fixes, bug fixes, features, breaking changes>
### Security
- <CVEs fixed, severity, whether the vulnerability affects this project's usage>
- If none: "No known security advisories in this version range"
### Compatibility
- Runtime: <compatible? evidence>
- Framework: <compatible? evidence>
- Cross-dependencies: <any conflicts with other pinned packages>
### Codebase impact
- <imports/usage found, any API changes affecting the codebase>
- Effort estimate: None / Low / Medium / High (with explanation)
### Recommendation
**Action**: Merge / Merge with caution / Hold / Close
**Urgency**: Critical (security) / High / Normal / Low
**Reasoning**: <1-3 sentences>
**Confidence**: <Confirmed / Likely / Speculative — and why>
After all individual PR sections, add:
## Cross-PR Analysis
- <dependency interactions between PRs>
- <recommended merge order, if any>
- <groups that should be merged together>
## Summary Table
| PR | Package | Action | Urgency | CI | Lockfile | Effort | Confidence |
|----|---------|--------|---------|----|----------|--------|------------|
## Manual testing suggestions
<If any PR warrants manual testing, describe what to look for>
model: "opus" for subagents. Research deeply.references/research-guide.md — Where to find changelogs, security advisories, compatibility information, and effective search strategies for dependency researchnpx claudepluginhub shanethacker/sst-claude-plugin --plugin sstGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.