cursor-review

Run a Cursor Composer review as an advisory second opinion while keeping the primary review verdict on the host brain. Use when user invokes cursor:review, asks Cursor to review, or wants composer to sanity-check a diff. Cursor never owns APPROVE/REQUEST_CHANGES.

Quality

85%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

82%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.

A tight, action-oriented skill body built around executable bash workflows with a clear sequenced flow and a real validation checkpoint. The main improvements are trimming the bilingual redundancy and either bundling the referenced scripts or signaling their external location.

Suggestions

Remove or collapse the Japanese opener since it duplicates the frontmatter description and adds tokens without new information.

Bundle the referenced scripts (cursor-companion.sh, resolve-impl-backend.sh, model-routing.sh) under scripts/ or add a one-line pointer to where they live, so the inline references resolve.

Add an explicit validate→fix→retry loop in step 4 (e.g., 'if Cursor flags a regression, re-run the diff check after the host applies a fix') to push workflow_clarity toward 5.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean, executable bash that assumes competence, but the Japanese opener ('Cursor Composer を read-only second opinion として使う…') restates the description bilingually and could be trimmed, so it sits at the efficient-but-not-fully-lean 4 rather than 5.	4 / 5
Actionability	Four complete, copy-paste-ready bash blocks (root resolution, scope parsing, Cursor prompt, host review) cover the common cases with real commands and explicit argument handling, matching the fully-executable 5 anchor.	5 / 5
Workflow Clarity	A clear four-step sequence with an explicit env-validation checkpoint (exit 2 on missing HARNESS_PLUGIN_ROOT); the operation is read-only so the destructive-cap does not apply, but there is no validate→fix→retry loop over Cursor's findings, leaving it just short of the 5 anchor.	4 / 5
Progressive Disclosure	Sections are well-organized with no nested references and the body is self-contained, but it references scripts (cursor-companion.sh, resolve-impl-backend.sh, model-routing.sh) that are not present in any bundle directory, a minor navigation gap that holds it at 4 rather than 5.	4 / 5
	Total	17 / 20 Passed

Description

87%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 strong description with explicit what/when structure, third-person voice, and a clearly scoped niche that avoids overlap with the primary review skill. The only gaps are slightly narrower action enumeration and trigger synonym coverage, which keep specificity and trigger_term_quality at 4 rather than 5.

Dimension	Reasoning	Score
Specificity	Names the domain and several concrete behaviors ('advisory second opinion', 'keeping the primary review verdict on the host brain', 'Cursor never owns APPROVE/REQUEST_CHANGES'), but the enumerated actions are narrow — all advisory/verdict-keeping rather than a broad action set, so it falls short of the comprehensive 5 anchor.	4 / 5
Completeness	Explicitly answers both 'what' (run advisory second-opinion review, keep verdict on host) and 'when' with concrete trigger phrases in a clear 'Use when…' clause, matching the 5 anchor.	5 / 5
Trigger Term Quality	Includes good natural triggers ('user invokes cursor:review', 'asks Cursor to review', 'wants composer to sanity-check a diff'), but coverage stops at these three phrases without the synonym spread the 5 anchor requires.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (Cursor advisory lane) with distinct triggers and explicit separation from the primary host verdict, giving minimal conflict risk per the 5 anchor.	5 / 5
	Total	18 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	15 / 16 Passed

Repository: Chachamaru127/claude-code-harness
Path: opencode/skills/cursor-review/SKILL.md
Commit: 7d8deba

Reviewed: about 15 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.