Skip to content
Code Review
Skill

/review-impl

Review uncommitted changes, unpushed or recent commits, or a branch diff against the plan/target with an external reviewer model (Codex CLI by default; Claude Code or Antigravity on request) — verify correctness, catch unacceptable deviations and bugs

From plugin
cross-review
1686 skills

Claude Code

$ npx -y skills add lploc94/cross-review --skill review-impl --agent claude-code

SKILL.md

review-impl.SKILL.md
name: review-impl
description: Review uncommitted changes, unpushed or recent commits, or a branch diff against the plan/target with an external reviewer model (Codex CLI by default; Claude Code or Antigravity on request) — verify correctness, catch unacceptable deviations and bugs (logic/security/memory/runtime). No new features or over-engineering beyond the plan. You apply valid fixes, rebut invalid points, and iterate until consensus or stalemate.

Implementation Review

Purpose

Verify the implementation matches the approved plan or target before commit or merge: what is missing, what deviates and whether that breaks the target, and real defects. Design and technology were settled in the plan; this review does not reopen them.

When to Use

After writing code: before committing, after committing and before pushing, before merging a branch, and at a plan's review checkpoints.

Inputs

  • **Mode**: `working-tree` (staged or unstaged changes exist) or `branch` (commits against a base; working tree clean). Detect with `git status --short` and `git rev-list @{u}..HEAD`. The diff must be non-empty.
  • **Base** (`BASE_BRANCH`, branch mode): the commit before the commits under review. Unpushed commits: `@{u}`. The last N commits: `HEAD~N`, with N no larger than `git rev-list --count HEAD` minus one; fix commits move `HEAD~N`, so resolve it once with `git rev-parse HEAD~N` and pass that SHA in every round. A branch or your own PR before merge: the target branch, with the PR title and description in `USER_REQUEST`. Validate it with `git rev-parse --verify <base>^{commit}`. A range that would include the repository's first commit has no base; tell the user the review would cover the whole project and ask how to proceed.
  • Round 1 placeholders: `USER_REQUEST` (original request plus the user's latest decisions; name the approved plan and phase when there is one), `SESSION_CONTEXT` (only what the reviewer cannot read in the plan, the diff, or the repository: decisions and constraints from the conversation, what was already verified; where real samples live when the work parses, limits or guards input), `BASE_BRANCH` (branch mode).
  • Reply placeholders: `SESSION_CONTEXT`, `FIXED_ITEMS`, `DISPUTED_ITEMS`, `BASE_BRANCH` (branch mode).
  • Skill `review-impl`.

{{PROTOCOL_CORE}}

{{PROTOCOL_VARIANT}}

For this skill: fix accepted issues in code, add a regression test where the defect is testable, and rerun the affected verification. Branch mode: commit the fixes as new commits before replying; never amend, rebase, or force-push reviewed commits unless the user asks. A `plan-gap` finding from the reviewer follows the plan-gap path above.

Completion

APPROVE: done. Stalemate: present the deadlocked issues and ask the user. Report rounds, verdict, issues found/fixed/disputed, user decisions, fixed defects by severity, and remaining risks.

Rules

  • If in plan mode, exit plan mode first; this skill edits code.
  • The reviewer only reviews and never edits files; you make the changes. Every accepted issue becomes a concrete code change.
  • Never hide review-originated scope expansion inside a bug fix, test adjustment, helper refactor, dependency change, or rebuttal.
Read more
@lploc94
Ships withcross-review

Cross-model review skills for Claude Code and Codex CLI: one agent orchestrates, another model reviews plans, code and reasoning.

Stars on GitHub

Momentum

stars in total
168
Star history of lploc94/cross-review on GitHub, a chart by star-history.com

Other skills on cross-review.

loop-implement

loop-implement

Execute an approved loop-plan/v1 plan end to end with docs-first TDD, verification, checkpoint reviews, and commits. Use only…

loop-plan

loop-plan

Create an execution-ready, code-grounded phased implementation plan before coding. Only for large work - several features, many…

review-plan

review-plan

Review/debate plans before implementation with an external reviewer model (Codex CLI by default; Claude Code or Antigravity on…

review-pr

review-pr

Review someone else's pull request or branch with an external reviewer model (Codex CLI by default; Claude Code or Antigravity on…

review-think

review-think

Peer debate with an external reviewer model (Codex CLI by default; Claude Code or Antigravity on request) on any technical…

More skills from other plugins.

open-code-review

open-code-review

Performs AI-powered code review on Git changes using the `ocr` CLI from alibaba/open-code-review. Use when the user asks to…

@alibaba@alibabaInstall Skill
build-graph

build-graph

Build or update the code review knowledge graph. Run this first to initialise, or let hooks keep it updated automatically.

@tirth8205@tirth8205Install Skill
hunk-extensions

hunk-extensions

Maps the `hunkdiff/extension` authoring surface for Hunk, the terminal diff viewer — hiding or reordering reviewed files, docked…

hunk6d9.8k
@modem-dev@modem-devInstall Skill

Questions about review-impl.

Which coding agents can use review-impl?
review-impl installs in Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and Windsurf with the skills CLI, using each agent's flag shown on this page. Flowy's auto-invocation, which fires a skill when your prompt matches, runs in Claude Code only.