Home/Skills/Select Review Method by Risk

Risk-Based Code Review Skill for Codex

Perform a read-only review of PR, commits, branches, or workspace differences, trace callers, verify requirements, and provide evidence-based findings. The default is an in-depth Solo review.

Tink1.1.0Nov 9, 2026MIT
Get the skill

Requirements

Intended for Codex-class agents that have repository and file access. The files do not provide browser, test tooling, credentials, project data, or permission to modify external systems; other hosts may require adaptation.

  • Clear baseline, target, and current differences
  • Repository rules, available requirements, tests, and relevant failure evidence
  • Host support for subagents is needed only when selecting independent multi-agent coverage

Quick start

  1. After download, first check the in-package documentation, then extract the ZIP to a temporary directory.
  2. When installing the combined set, copy the six Skill folders and the sibling shared folder into the project's .agents/skills/ or personal ~/.agents/skills/. Keep the relative directory structure, and do not treat the outer combined-set directory as one Skill.
  3. When installing an individual ZIP, put its identically named folder into the selected Skills directory and keep its internal shared resources. Back up the same-named Skill before overwriting to avoid duplicate installation.
  4. In Codex, use the exact $code-review-multiagent invocation; if not found, restart the host. When a task requires workflow handoff, install the corresponding companion Skill separately.

Applicable Scenarios

  • Review changes pending merge against requirements and direct callers.
  • Complex changes involving security, data, interface contracts, or tests with independently investigable surfaces.

Usage Steps

  1. Call $code-review-multiagent to clarify the baseline, target, and requirement source.
  2. Trace the direct callers of the modified public symbols, check applicable risks, and also check behaviors required by requirements but completely missing from the diff.
  3. Default to Solo; only assign 2–4 independent review perspectives when complexity truly warrants and host support is available.
  4. Verify and merge findings with the same root cause, and report severity, trigger conditions, location, impact, evidence, and minimal fix direction.

Deliverables

  • Prioritized P0–P3 findings, or a scoped "no provable defect found" conclusion
  • Provide normative, requirement, and runtime acceptance status and validation gaps separately.

Limitations

  • Does not automatically fix, and does not list pure style preferences as defects.
  • Cannot claim that the candidate implementation was reviewed when there is no diff.
  • Static review and passing tests cannot prove browser usability or deployment; missing runtime evidence must be explicitly preserved.

Example Prompts, Not an Actual Execution

Use $code-review-multiagent to review the current workspace differences against the main branch and approved access control requirements. Keep it read-only, trace direct callers and side effects of rejected requests; distinguish confirmed issues from missing runtime evidence.

This guide describes the in-package workflow and does not represent verified cross-model reliability, nor is it an actual acceptance run report.

Engineering standards and requirements

Review the latest diff along actual callers and user paths. Report findings and verification gaps separately for standards and the requirements. The ZIP includes a Fowler code smell reference for evaluating concrete maintenance and correctness issues.

Inside the skill
SKILL.md
---
name: code-review-multiagent
license: MIT
description: >-
  Review a PR, commit, branch, or working-tree diff read-only. Trace callers, check engineering standards and requirements separately, and report evidence-based severity. Use one reviewer by default and split independent complex surfaces when needed.
---

# Code Review Against Standards and Requirements

Trace requirements, actual callers, and concrete failure conditions. Keep review read-only. Authorized reads and checks without side effects are allowed. External cost, sending, or other writes require applicable authorization.

## Set the scope

- Use the specified baseline and paths first. Use the task starting point for “this change” when established. Ask only for missing scope when unrelated changes cannot be separated.
- Confirm the reference resolves and the diff is nonempty. For commits use git diff base...tip and record the commit list. For the working tree use git diff base with path limits. Use git diff --cached for staged changes. Read in-scope untracked files separately.
- Use the user's explicit requirements and specified spec, then associated issues or existing specs. A chat requirement is valid. If there is no spec, mark that gap and finish the standards review. An agent's own plan is not user approval.
Included files · 10
code-review-multiagent/LICENSE
code-review-multiagent/MANIFEST.json
code-review-multiagent/README.md
code-review-multiagent/SKILL.md
code-review-multiagent/agents/face-reviewer.md
code-review-multiagent/agents/finding-challenger.md
code-review-multiagent/agents/openai.yaml
code-review-multiagent/references/code-smells.md
…