sdlc-doubt SKILL.md skillA
sdlc-doubt from tscarpe/claude-sdlc-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add tscarpe/claude-sdlc-skills/sdlc-doubt@git:20261018.62c92c0Pins git:20261018.62c92c0 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- tscarpe/claude-sdlc-skills
- Path
skills/sdlc-doubt/SKILL.md- Versions recorded
- 7
- Latest version
- git:20261018.62c92c0, committed Oct 18, 2026
- First version recorded
- Sep 12, 2026
- Size
- 6,820 bytes
- Repository stars
- 66
- License
- MIT
Adversarially re-checks a non-trivial technical decision in a fresh context by stripping the conclusion and asking an independent reviewer to find problems. 对刚做出的非平凡技术决策发起对抗式独立复查——把决策剥离结论后交给全新上下文的审查者"找问题"。Use when 影响ER或公共组件的决策回写前、复杂SQL/状态机流转定稿前、技术方案选型二选一定稿前、提交前对"已全范围检查"断言存疑时、用户说"这个判断我不放心""帮我质疑一下这个方案""这个方案稳不稳""doubt-review"。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Doubt Review（对抗式决策复查） > **定位**：自信的答案不等于正确的答案。长会话中，假设会悄然变成"事实"。本 skill 是 **in-flight** 复查——决策落地前质疑它，不是事后的代码质量检查（如 trellis-check、/code-review）。 **前置依赖**：Agent 工具（生成全新上下文子代理）；无脚本或外部服务依赖。 ## 何时触发 ### 非平凡决策定义（任一命中才触发） 1. 新增/修改分支逻辑或状态机流转 2. 跨模块/跨域边界，影响 ER 或公共组件 3. 断言编译器无法验证的属性（事务边界、幂等、数据一致性、租户隔离） 4. 不可逆或大爆炸半径（删数据、迁移、审核状态变更） ### 触发场景 - 大需求切片发现影响 ER/公共组件的偏差，**回写上层设计文档前** - design.md 关键决策定稿前（技术设计评审前） - 复杂 SQL 定稿前（join 3 表+ / 租户隔离条件 / 聚合口径） - 代码提交前，对"全范围检查已通过"**这一断言本身**存疑时 - 用户明示触发 分级关系（读梳理文档头「风险分级」，缺失时按非平凡决策定义本身判定）：A 级 = 命中非平凡决策定义即**强制**触发；B 级 = 可选（用户可跳过）；C 级 = 跳过（仅用户明示时触发）。 ### 不触发（防滥用） 命名选择、单层内实现细节、spec 已直接覆盖的规则、机械性改动（重命名/格式化/移动文件）、用户明确要速度不要验证。 ## 五步闭环 复制此 checklist 跟踪： ``` Doubt cycle: - [ ] Step 1: CLAIM — 2-3 行命名决策 + 为何重要 - [ ] Step 2: EXTRACT — 提取 artifact + contract，剥离推理，不含 CLAIM - [ ] Step 3: DOUBT — Agent 子代理 + 对抗 prompt - [ ] Step 4: RECONCILE — 每条发现按四分类回 artifact 文本归类 - [ ] Step 5: STOP — 满足停止条件（平凡发现 / 3 轮 / 用户 override） ``` ### Step 1: CLAIM ``` CLAIM: "订单取消走快照回滚而非实时表反查，在并发操作场景下数据一致" WHY: 不一致会导致取消后订单列表与操作记录不符 ``` 写不出 2-3 行的紧凑 CLAIM = 还没有决策，只有感觉——先想清楚再质疑。 ### Step 2: EXTRACT（最关键的一步） 给审查者只准备两样东西： - **ARTIFACT**：被审制品——代码 diff、design 片段、SQL，**不是整个文件** …
Read the whole file at its exact version.
How to install
mdr add tscarpe/claude-sdlc-skills/sdlc-doubt@sha256:7ea87cb453ad7c1eThe line at the top of this page pins the label git:20261018.62c92c0, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_wslsamspfjc5bmkl)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 7 versions of this file, between Sep 12, 2026 and Oct 18, 2026, about one every 6 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20261018.62c92c0 latest | 2026-10-18 | 62c92c0 | 6,820 B | A | view · diff |
| git:20260930.29f1d9d | 2026-09-30 | 29f1d9d | 6,558 B | A | view · diff |
| git:20260929.dd14fd1 | 2026-09-29 | dd14fd1 | 6,211 B | A | view · diff |
| git:20260928.9ef9188 | 2026-09-28 | 9ef9188 | 6,168 B | A | view · diff |
| git:20260926.fa9dd0b | 2026-09-26 | fa9dd0b | 6,166 B | A | view · diff |
| git:20260916.e9908db | 2026-09-16 | e9908db | 5,917 B | A | view · diff |
| git:20260912.83789e8 | 2026-09-12 | 83789e8 | 5,819 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (6820 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
tscarpe/claude-sdlc-skills · 66 stars · license MIT · pushed 2026-11-05 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_wslsamspfjc5bmkl GET https://markdownregistry.com/api/v1/resolve?ref=tscarpe/claude-sdlc-skills/sdlc-doubt GET https://markdownregistry.com/api/v1/blob/7ea87cb453ad7c1ebbee677b23e44c38253a81a4160deed5588e56c0fe8aca13
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.