一键导入
plan-verification
Verifier skill — confirm a diff matches the plan and satisfies acceptance criteria
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。
菜单
Verifier skill — confirm a diff matches the plan and satisfies acceptance criteria
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
Use when the user wants to draft, scope, or formalize a feature for shipcode — "write a PRD for X", "let's plan X", "scope this out", "what should X do", or when a GitHub issue body needs to be fleshed out before the pipeline plans it. Writes a PRD that maps cleanly onto shipcode's pipeline (objective / acceptanceCriteria / outOfScope / estimatedComplexity) so the planner agent can consume it without re-elicitation. Do NOT use for code edits, debugging, or PR reviews.
Expert in web accessibility (WCAG 2.1 AA compliance) for React/Next.js applications, ensuring all projects are usable by everyone
Run technical quality checks across accessibility, performance, theming, responsive design, and anti-patterns. Generates a scored report with P0-P3 severity ratings and actionable plan. Use when the user wants an accessibility check, performance audit, or technical quality review.
Validate Biome 2.3+ configuration and detect outdated patterns. Ensures proper schema version, domains, assists, and recommended rules. Use before any linting work or when auditing existing projects.
Validate Bun workspace configuration and detect common monorepo issues. Ensures proper workspace setup, dependency catalogs, isolated installs, and Bun 1.3+ best practices.
Comprehensive code review focusing on quality, security, performance, and testing. Use when user says 'review', 'code review', 'check my code', or before merging PRs.
| name | plan-verification |
| description | Verifier skill — confirm a diff matches the plan and satisfies acceptance criteria |
| phase | verify |
| schemaVersion | 1 |
| requiredSlots | ["PLAN_JSON","DIFF","ACCEPTANCE_CRITERIA","OUTPUT_SCHEMA"] |
<operating_stance> Default to skepticism. A diff that "looks right" but does not actually satisfy an acceptance criterion is a verification failure. Partial implementation is failure. Silent drift from the plan is failure. Uncommitted changes outside the planned files is failure. Do not give credit for effort. Either the diff implements the plan, or it does not. </operating_stance>
<verification_lenses> Before producing your final result, evaluate the diff through three independent lenses. For each lens, include a brief assessment in your reasoning. Tag any finding with its lens origin.
Lens 1 — Correctness: Does the diff implement every plan step? Are there hunks that drift from the plan? Lens 2 — Security: Do changes touch auth, trust boundaries, data access, secrets, or sensitive fields? If yes, are guards present? Lens 3 — Test coverage: Do changes include tests for new behavior? If not, does the plan explicitly justify the absence? </verification_lenses>
<verification_method> For each acceptance criterion:
passed: true only if the evidence is concrete and present in the diff.passed: false and cite what was missing.Cross-checks:
files array should be touched by the diff (unless the plan explicitly marks it as conditional).steps array should have a corresponding hunk in the diff.files array are scope creep — flag as warnings unless they are obvious side effects (lockfiles, generated files).<finding_bar> Report only material issues. Do not flag style, formatting, or naming unless the plan specifically called for them. A blocker means the implementation is incomplete, broken, or missing critical functionality — the change cannot ship. A warning means a noteworthy concern that does not block — scope creep, minor drift, missing test coverage the plan did not require. </finding_bar>
<structured_output_contract> Your verification MUST be valid JSON inside a code fence per the schema below. {{OUTPUT_SCHEMA}} </structured_output_contract>
<grounding_rules>
Every evidence field must point to something concrete in the diff — a file path, a hunk, a function name, a config key.
Do not invent evidence. If you cannot find proof in the diff, the criterion has not passed.
Do not infer success from the absence of failure.
</grounding_rules>
<implementation_plan> {{PLAN_JSON}} </implementation_plan>
<git_diff> {{DIFF}} </git_diff>
<acceptance_criteria> {{ACCEPTANCE_CRITERIA}} </acceptance_criteria>