Name: review-pr
Rating: 67.2 (1 reviews)
Author: nrwl

review-pr

Deep code review of a single open PR in nrwl/nx. Checks the PR out only inside an isolated sandbox, then runs four fixed reviewers: implementation (correctness, errors, types, performance), verification (tests, ticket grounding, comments, and docs), approach, and security. A reproduce-verifier executes a runnable repro only when verification identifies one. The skill saves a GitHub-flavored draft to ~/.nx-pr-reviews/<NUMBER>.md and never posts it. Claude reads/executes PR code only through the sandbox CLI; credentials never enter the sandbox.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

An exceptionally detailed, executable multi-step workflow with strong validation checkpoints and real bundle references — actionability and workflow clarity are exemplary. The chief weakness is conciseness: extensive historical-rationale prose inflates a 1699-line single file that could offload templates to reference files.

Suggestions

Trim the recurring 'Observed: ...' historical-bug anecdotes to a terse one-line justification or move them to a NOTES reference file; they explain past failures but cost tokens on every load.

Extract the large charter and dispatch-prompt templates (Steps 5 / 5a / 5a.3 / 5a.5) into a references/ template file the body links to, reducing the inline SKILL.md footprint.

Consolidate the repeated sandbox-protocol explanations (the same read/grep/find/exec verbs and host-execution warnings appear in multiple steps) into one referenced section.

Dimension	Reasoning	Score
Conciseness	Most prose is load-bearing and assumes Claude's competence, but the 1699-line body repeatedly over-explains rationale and historical bugs ('Observed: an image predating the pnpm-store warming went unnoticed for two weeks', 'Observed on a single PR: six separate harnesses...') that pad token cost beyond the operational instruction.	3 / 5
Actionability	Every step ships concrete, executable bash — e.g. '.claude/tools/sandbox start --image "$SANDBOX_IMAGE" --checkout ... --ref pull/<NUMBER>/head --base <BASE_REF_NAME>', the gh JSON calls, and the full EVIDENCE-verification case block — copy-paste ready and covering the common cases.	5 / 5
Workflow Clarity	Steps 1–10 are explicitly numbered and sequenced with validation checkpoints throughout: write-verify-move on the diff, EVIDENCE proof-of-work gating with retry, admission-test gates, and the destructive/batch feedback loops (force-push fallback, sandbox stop in Step 9 even on failure) that the rubric specifically rewards.	5 / 5
Progressive Disclosure	Structure is good: a clear stepwise overview with one well-signaled one-level-deep reference to scripts/replay-prior-patch.sh (a real bundle file). It is not a 5 because large per-run template blocks (the charter, the dispatch prompts) are inlined rather than split into reference files, and the skill leans on a monolithic SKILL.md.	4 / 5
	Total	17 / 20 Passed

Description

83%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A precise, third-person description that concretely enumerates the skill's mechanics and trust model with minimal fluff. Its main weakness is the absence of an explicit 'Use when...' trigger clause, leaving the 'when' only weakly implied.

Suggestions

Add an explicit 'Use when reviewing an open PR in nrwl/nx' clause so the trigger is stated, not implied.

Include natural synonyms like 'pull request' alongside 'PR' to broaden trigger-term coverage.

Name the skill's argument shape (e.g. 'given a PR number') in the description to make invocation discoverable from the frontmatter alone.

Dimension	Reasoning	Score
Specificity	Lists many concrete actions — 'Checks the PR out only inside an isolated sandbox', 'runs four fixed reviewers: implementation...verification...approach...security', 'A reproduce-verifier executes a runnable repro', 'saves a GitHub-flavored draft to ~/.nx-pr-reviews/<NUMBER>.md and never posts it' — comprehensive coverage of the skill's mechanics.	5 / 5
Completeness	It clearly answers 'what' (sandboxed checkout + four reviewers + reproduce-verifier + draft) but the 'when' is only implied by 'a single open PR' and the argument-hint; there is no explicit 'Use when...' clause, which per the guidelines caps it below 5.	4 / 5
Trigger Term Quality	Natural terms like 'code review', 'open PR', 'nrwl/nx', 'reproduce-verifier', and 'repro' are present and what a user would say, but it lacks synonyms/variations such as 'pull request', 'diff review', or explicit file-extension-less phrasing that would push it to 5.	4 / 5
Distinctiveness Conflict Risk	The nrwl/nx-specific scope, the isolated-sandbox trust model, and the named fixed reviewers carve a clear niche unlikely to trigger for any other skill; conflict risk is minimal.	5 / 5
	Total	18 / 20 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (1700 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: nrwl/nx
Path: .claude/skills/review-pr/SKILL.md
Commit: 9064aa9

Reviewed: about 4 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.