CtrlK
BlogDocsLog inGet started
Tessl Logo

finsi/codex-review

Use when the user wants a local second-opinion code review via OpenAI Codex CLI — on the current branch, staged changes, a single file, or a piped diff. Triggers include "codex review", "review with codex", "run codex on this diff", "second opinion from codex", or pre-commit / pre-PR review requests that ask for codex specifically.

90

2.25x
Quality

97%

Does it follow best practices?

Impact

97%

2.25x

Average score across 2 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

Quality

Content

92%

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

This is a high-quality, well-crafted skill that provides clear, actionable guidance for running Codex CLI reviews. The workflow is well-sequenced with appropriate gates and validation steps, and the content is impressively concise while covering edge cases and common pitfalls. The main weakness is the dependency on PROMPT.md which isn't provided in the bundle, making the skill incomplete as a standalone artifact.

Suggestions

Include the referenced PROMPT.md in the bundle or inline its key invariants so the skill is self-contained and the reference can be verified.

DimensionReasoningScore

Conciseness

The skill is lean and efficient throughout. It assumes Claude's competence with git, shell commands, and CLI tools. Every section earns its place — no explanations of what diffs are, how git works, or what severity levels mean. The tables and inline notes are tightly written.

3 / 3

Actionability

Fully executable bash commands throughout, with exact flags and file paths. The scope table gives copy-paste git commands, the canonical invocation is complete, and even edge cases (single file, Claude Code shell) have concrete command examples. The Common Mistakes table provides specific fixes.

3 / 3

Workflow Clarity

Clear 5-step numbered workflow with explicit gates (preflight checks, `test -s` guard is marked mandatory), validation checkpoints (verify CRITICAL/HIGH against actual code before editing), and feedback loops (check test suite before applying fixes). The triage step explicitly warns about hallucinated line numbers and deliberate design choices.

3 / 3

Progressive Disclosure

The skill references `.claude/skills/codex-review/resources/PROMPT.md` multiple times as a critical dependency, but no bundle files are provided, making it impossible to verify the reference or understand what invariants it encodes. The content itself is well-structured with clear sections, but the missing PROMPT.md is a significant gap since the skill explicitly states it's 'what makes it useful.'

2 / 3

Total

11

/

12

Passed

Description

100%

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

This is an excellent skill description that clearly communicates what the skill does (local code review via Codex CLI producing severity-grouped findings), when to use it (explicit trigger phrases and use-case scenarios), and how it's distinct from other skills (Codex-specific triggers). The description is information-dense without being padded, uses third-person voice appropriately, and provides strong disambiguation from generic code review skills.

DimensionReasoningScore

Specificity

Lists multiple specific concrete actions: severity-grouped findings (CRITICAL/HIGH/MEDIUM/LOW), file:line locations, one-line fixes, scoping to branch/staged/file/diff, and enrichment with repo invariants (multi-tenant/auth/migration).

3 / 3

Completeness

Clearly answers both 'what' (second-opinion code review producing severity-grouped findings with locations and fixes) and 'when' (explicit 'Use when' clause at the start plus a 'Triggers include' clause listing specific phrases).

3 / 3

Trigger Term Quality

Includes highly natural trigger phrases users would actually say: 'codex review', 'review with codex', 'run codex on this diff', 'second opinion from codex', plus contextual triggers like 'pre-commit / pre-PR review requests'. Good coverage of variations.

3 / 3

Distinctiveness Conflict Risk

Highly distinctive — specifically scoped to OpenAI Codex CLI (gpt-5.5) as the review tool, with explicit trigger terms mentioning 'Codex' that clearly distinguish it from generic code review skills or other review tools.

3 / 3

Total

12

/

12

Passed

Validation

100%

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

Validation11 / 11 Passed

Validation for skill structure

No warnings or errors.

Reviewed

Table of Contents