verify-this
Formal proof of one concrete claim. Use only for an explicit verify, prove, measure, or confirm request, or when a required high-risk acceptance gate needs a verdict.
- License 许可
- MIT
- 最近更新
- 2026-11-14
- Stars
- ★ 2
- Forks
- 0
01 前言 Front Matter
- name
- verify-this
- source
- https://github.com/shaneholloman/cursor-plugins/tree/main/cursor-team-kit/skills/verify-this
- license
- MIT
- description
- Formal proof of one concrete claim. Use only for an explicit verify, prove, measure, or confirm request, or when a required high-risk acceptance gate needs a verdict.
02 资源文件 1
- 参考文档agents/openai.yaml225 B
03 SKILL.md 正文
Verify This
Use only for an explicit verification request or a required high-risk acceptance gate. Goal: prove or disprove one falsifiable statement. Not for routine completion checks, ordinary test/lint runs, general research, or vague quality claims ("the code is cleaner").
Use GO/NO-GO mode for acceptance gates: high-risk completion checks, migration readiness, PR cleanup readiness, or independent verifier decisions.
Invoke once per claim. Do not reread this skill or restart the workflow in the same task unless the claim, threshold, evidence, or environment changed.
Workflow
- Restate claim in falsifiable form: condition, metric, threshold, expected direction.
- Capture baseline evidence when possible (merge base, parent commit, current broken repro), then treatment evidence with the same command, data, and environment. Compare raw artifacts, not summaries.
- Return exactly one verdict.
Prefer inline evidence. Do not persist secrets, credentials, customer data, or private code extracts without explicit consent.
Verdict Rules
VERIFIED: baseline and treatment differ in predicted direction by claimed threshold with no obvious confound.NOT VERIFIED: behavior unchanged, moves wrong way, or misses threshold.INCONCLUSIVE: no valid baseline, noisy signal, failed measurement, or environment difference invalidates comparison.
VERIFIED | NOT VERIFIED | INCONCLUSIVE
Claim: <falsifiable claim>
Evidence:
<metric/artifact>: baseline=<...>, treatment=<...>, delta=<...>, threshold=<...>
Reasoning:
<one tight paragraph naming evidence and confounds>
GO/NO-GO Mode
For shipping decisions rather than a single claim verdict.
- Restate acceptance criteria as falsifiable checks.
- Verify against current local artifacts/code/live UI. Do not trust prior assistant reports.
- Check every expected artifact exists, is non-empty, matches source of truth; check no secrets or sensitive data persisted when relevant.
GOonly when every required check passes with current evidence.NO-GOwhen any check fails, is blocked, or cannot be verified.
GO | NO-GO
Scope: <what was verified>
Evidence:
| Check | Evidence | Result |
| --- | --- | --- |
| <criterion> | <artifact/command/live check> | PASS/FAIL |
Repair actions:
- <only when NO-GO>
Residual risks:
- <only real remaining risks>
Do not soften negative results. Clear NOT VERIFIED is useful.
04 同仓库更多 Skills
writing-plans
Use when you have a spec or requirements for a multi-step task, before touching code
workflow-from-chats
Extract preferences from agent chats → skills, rules, workflow docs. For learning preferences, mining feedback, personalizing workflows.
winui3-csharp-app
WinUI 3 / Windows App SDK desktop apps in C#/XAML. MVVM, bindings, navigation, Mica/Acrylic, windowing, packaging, performance.
why
Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available MCPs and queries each evidence category (source control, issue tracker, long-form docs, real-time chat, infrastructure observability, error tracking, product analytics warehouse) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.
what-did-i-get-done
Summarize authored commits over a user-specified time period into a concise status update.
weekly-review
Produce a short weekly synthesis of authored commits with highlights by bugfix, tech debt, and net-new work.