lfx-review-pr

Reviews GitHub pull requests against LFX architecture standards and repo conventions.

12|6|Updated Jul 14, 2025
One-click install
npx skills add https://github.com/linuxfoundation/lfx-self-serve --skill lfx-review-pr-linuxfoundation
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: lfx-review-pr
Source: https://github.com/linuxfoundation/lfx-self-serve/tree/main/.claude/skills/lfx-review-pr
Command: npx skills add https://github.com/linuxfoundation/lfx-self-serve --skill lfx-review-pr-linuxfoundation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reviewing pull requests against a large set of repo conventions, architecture rules, and prior reviewer feedback is slow and error-prone. This Skill automates the full PR audit pipeline so reviewers get a complete, verified draft review without manually checking every rule. ## Core Features & Use Cases - Parallel code audits: Launches two reviewer subagents (general code review and self-serve convention/architecture review) that analyze the PR branch's diff against origin/main and return markdown reports. - Prior comment verification: Fetches previous review comments via the GitHub API and verifies against the actual code whether each CRITICAL or SHOULD_FIX item was fixed, partially fixed, or ignored. - PR-shape checks: Validates branch naming, ticket references, conventional commits, DCO sign-offs, GPG signatures, rebase status, diff size, protected files, PR title format, and external upstream references. - Safe posting workflow: Compiles everything into a draft review, presents it for explicit user approval, and only then posts via the /review skill — it never auto-posts. - Use Case: A maintainer runs the skill on PR #142; it detects that a prior comment about missing license headers is still unresolved, flags a non-conventional commit message, and drafts inline comments with rule citations for a first-time contributor. ## Quick Start Ask the assistant to review pull request 142 against the LFX standards after checking out the PR branch with gh pr checkout.

Frequently Asked Questions about lfx-review-pr

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a GitHub pull request against repo conventions automatically?▼

Check out the PR branch with gh pr checkout, then invoke the skill with the PR number. It fetches PR metadata, launches two reviewer subagents against origin/main, verifies prior comments, checks PR-shape rules, and presents a draft review for your approval before posting.

How does the skill verify that previous review comments were addressed?▼

It fetches inline comments and review bodies via the GitHub API, skips trivial or bot comments, then reads the current code on the PR branch using git show on the fetched pull ref. Each CRITICAL or SHOULD_FIX comment is classified as FIXED, NOT FIXED, PARTIALLY FIXED, or N/A in a markdown table.

Does the skill post review comments to GitHub automatically?▼

No. It never auto-posts comments or submits reviews. It always presents the full draft — inline comments, review body, and verdict — and waits for explicit user approval through an interactive prompt before invoking the /review skill.

What PR-shape checks does the review include?▼

It validates branch naming, ticket references, conventional commit subjects, PR title format, rebase status against the base ref, diff size, DCO Signed-off-by trailers, GPG signatures, protected files, and external upstream references for server-side changes.

Why does the skill fail with a HEAD mismatch error?▼

The skill requires your local HEAD to match the PR's tip commit so the diff against origin/main is accurate. If the check fails, commit or stash your work and run gh pr checkout with the PR number, then restart the skill.

Can the review adapt its tone for first-time contributors?▼

Yes. It checks how many merged PRs the author has in the repo; authors with fewer than five trigger educational mode, where inline comments explain the reasoning behind each rule and cite the exact rule file and section.