원클릭으로
review-implementation
Review code and prose changes against the plan they implemented.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Review code and prose changes against the plan they implemented.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Rebase the current branch with smart conflict resolution.
Manage, organize, sort, rename, dedupe, or archive files on disk. Use when reshaping a directory tree, batch-renaming, deduplicating, sorting media by date or location, mounting or extracting disk images, or cleaning up a folder. Skip for git operations and code edits.
Guide for creating Dagger modules, toolchains, and CI pipelines using the Go SDK. Use this skill whenever the user wants to create or modify a Dagger module, build a toolchain, write Dagger Functions in Go, or set up CI/CD with Dagger. Also trigger when you see dagger.json, the dagger CLI, imports from dagger, or references to the `dag` client.
Reference and guidance for writing Taskfiles (Taskfile.yaml) and using the `task` / go-task CLI (https://taskfile.dev/). Trigger when the user edits or creates Taskfile.yaml files, runs or troubleshoots the `task` CLI, or asks about Task or `go-task`. SKIP when the user is discussing generic to-do lists, Celery/Airflow/Luigi tasks, Claude Code's own TaskCreate/TaskList tools, or macOS Reminders.
Retrospective on the current session's friction (broken tool calls, repeated prompts, wasted effort), with proposed fixes.
Launch one or more tasks in new git worktrees using workmux.
| name | review-implementation |
| description | Review code and prose changes against the plan they implemented. |
Review all changes made during plan implementation against the plan itself. Address any issues found.
The caller provides the plan file path and the base SHA (commit hash from
before implementation began). Run git diff <base-sha> to see the full diff
(committed and uncommitted) since implementation began.
Use the Agent tool to launch both agents concurrently in a single message. Both agents read the full diff; they split by concern, not by file. Pass each agent the plan file path, the base SHA, and any specific concerns the user mentioned.
Launch implementation-reviewer-code to review correctness across all
changed files: logic, completeness against the plan, deviations, tests,
simplicity, security, and behavioral correctness.
Launch implementation-reviewer-docs to review prose quality across all
changed files: self-containment, timeless framing, unnecessary comments,
docs completeness against the plan, and compliance with project conventions.
Applied to in-code comments, docstrings, and standalone docs alike.
Wait for both agents to complete. Aggregate their findings and fix each issue directly. If a finding is a false positive or not worth addressing, note it and move on -- do not argue with the finding, just skip it.
After fixing, re-run both agents in parallel. Continue until both return
LGTM!. When done, briefly summarize what was fixed.