菜单
Fixed four-step plan review pipeline: plan-review (requirements validation mode), plan-review (interactive mode), product-review, review summary to Kestral. Orchestrates the full review pass for an implementation plan. Only invoked by the kstack subagent.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Validate the branch diff against the task's acceptance criteria before marking it done. Use when implementation looks complete, before a status update to done/awaiting_review, or when the user asks "did we cover everything?"
One-pass AI code review of a PR or branch diff (tech-lead style). Use when the user asks for a code review, PR review, AI review, or wants code quality, architecture, bugs, and patterns reviewed.
Simplify changed code in a PR or branch for readability while preserving behavior. Use when the user asks to simplify code, reduce complexity, clean up a PR, or improve readability of recent changes.
Capture new project-specific learnings into context files. Runs during reviews or after debugging to evolve the project's skill knowledge base. Use when the thread reveals patterns, anti-patterns, conventions, or codebase gotchas that should be remembered.
Debug pipeline: investigate (prod or local) → fix-issues → scoped polish (deslop + code-simplify) → pattern-check → kestral-sync retroactive capture. Only invoked by the kstack subagent.
Capture the decision when a spike, prototype, or investigation concludes. Use when a prototype or spike wraps up, the user makes a go/no-go call after de-risking, or an investigation answers its question — prompts "what did you decide?" and writes the decision + rationale to the task tracker.
| name | review-plan-pipeline |
| kstack | true |
| description | Fixed four-step plan review pipeline: plan-review (requirements validation mode), plan-review (interactive mode), product-review, review summary to Kestral. Orchestrates the full review pass for an implementation plan. Only invoked by the kstack subagent. |
| disable-model-invocation | true |
You are reviewing an existing implementation plan to production-ready quality. Execute four steps in strict sequence. Do not implement code — only review, validate, and surface issues for the plan.
Start by creating a todolist with one item per step below. Each item's first sub-action is reading that step's SKILL.md via the Read tool. Mark an item in_progress only after the Read tool call succeeds.
Read .agents/skills/plan-review/SKILL.md in full. Use Mode 2: Requirements Validation. Treat the plan as the
requirements document. Extract atomic requirements, verify each against the codebase, and classify as implemented /
partial / missing / unclear / needs manual verification.
Present the validation report to the user. Flag any high-risk gaps.
Read .agents/skills/plan-review/SKILL.md — use Mode 1: Interactive Plan Review (which references
review-process.md). Perform an interactive, section-by-section review of the plan against the codebase. Surface issues
and pause for user input between sections.
Read .agents/skills/product-review/SKILL.md in full (including its review-process.md reference) and follow its
workflow. Review the plan for surface area completeness — ensure every place the feature appears in the product is
accounted for, multi-user collaboration scenarios are considered, and AI agent interaction points are covered.
If a task is linked to the plan (task ID in frontmatter, or lookup finds one), post a Review Summary (see context.md for project-specific sync method). Scope: the plan file reviewed; findings: validation gaps + review issues; verdict: ready to implement or needs revision. Skip only if no task is linked.
After all steps, present a final summary:
.cursor/rules/ (read engineering-principles.mdc and meta-workflow.mdc).Read context.md and apply it as additional project-specific constraints layered on top of this
workflow. If it does not exist, skip this section.