second-take SKILL.md skillA
second-take from flashfrogluo/second-take is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add flashfrogluo/second-take/second-take@v1.7.4Pins v1.7.4 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- flashfrogluo/second-take
- Path
SKILL.md- Versions recorded
- 9
- Latest version
- v1.7.4, committed Oct 26, 2026
- First version recorded
- Oct 12, 2026
- Size
- 52,626 bytes
- Repository stars
- 3
- License
- MIT
Use when an AI answer is **bad / wrong / unusable** and you want to **fix or review** it, or for **prompt QA** (AI reasoning / deep thinking review)：诊断那段深度思考（推理链 / thinking）哪里出错，并产出一段可粘回原对话的重拍单（retake note）。**不适用**：从零创作新内容、纯信息查询（无需诊断已有推理）、闲聊。覆盖文本推理与图像 / 视频生成两类场景；生成平台出图出片不对时，也可用它质检提示词。完整工作流：重拍 → 补拍 → 定剪 → 调色 → 包装。**默认给轻装版（多为定点修复：只改几处；要完整版说「详细点 / 往深了查」）。** 用户说「这个结果不对 / 很难用 / 漏了要点 / 字数不达标 / 输出不像人话 / AI 味太重 / 格式不对」，或贴来分享链接、一段深度思考、一份 rubric要求质检，或点名「重拍 / 补拍 / 定剪 / 调色 / 包装」「帮我取舍 / 润色 / 可视化 /…
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
# Second Take · 再来一条 AI 深度思考质检与重拍单生成。 ## 轻装出口（默认行为） 本 skill 默认给**轻装版**：诊断照旧满配，交付默认精简。先按缺陷决定修改范围，再按复杂度决定展示篇幅。主体成立时做定点修复；主体不成立或用户明确要求重写时全量重写。首次审查也可以只改几处，多轮迭代也可能需要重拍；轮次本身不决定范围。 - **想照旧**：仍认真读完那段深度思考，仍用缺陷分类（6 检查项 / 13 子类）逐个过——这一步一点不能省，它就是这 skill 的本事。 - **写精简**：默认只给「一两行说明 + 约 7 行能直接粘的重拍单」，不写完整仪式。 **走短版**：用户说清了哪里不对，或明显只一处问题。 **走完整版（= 本 skill 现有全部流程，无需改动）**：用户说「重要 / 详细点 / 往深了查」，或问题明显结构性（多处不对、方向错、已改好几轮还不行）。 **首句按修改范围二选一**（只把选中的那句写进模板开头，不要把这段选择说明粘回原对话）： - 全量重写（主体不成立或用户明确要求）：舍弃已确认错误的推理与不合格结构，按用户要求重写；用户指定的保留项继续有效。 - 定点修复（上一版主体已成立，只改几处）：沿用上一版中你认可的部分（结构、合格段落、设定继续有效）；只收回上一轮那段深度思考里错误的取舍。你此前提出的要求继续有效。 短版模板（贴回原对话即可；首句用上面选中的那句，其余照抄）： > <首句：按修改范围选中的那句> > 任务：<用户指令全文> > 本次要改的：<缺陷逐条列出>——按选定范围执行：全量重写仍保留用户指定项；定点修复未点名处保持原样。 > 只输出用户要求的内容，不附加分析过程或答题旁白；用户要求的表格、清单、JSON、代码和保留表达优先于模板建议。 > > 输出前自答，两条都为「是」再输出（自答过程不写进正文）： > 1. 这次点名的每处修改是否都已落实，正文里已不再出现被点名的问题？ > 2. 定点修复时：未点名的地方是否保持原样？全量重写时：用户明确指定的保留项是否仍被保留？ （第 1 条防「改了没改到位」——静默白跑一轮；第 2 条（定点修复时）防「定点修复时把上版合格内容一起丢掉 / 版本退步」——后者是踩过最多次的头号教训，硬约束 15 整条都在讲它。两条查的都是「修改本身」，与产物长短无关，故任何档位都留（第 2 条在全量重写下改查用户指定保留项是否保留）。） **【判定标准】短版默认不加**；仅当用户指令自带领域硬约束（「符合学术规范」「按品牌调性」「要过审」）时，加一条：「这条修改有没有本领域的隐性验收标准？有就列 1 条并对照，没有就答『无』。」一行成本，保住事前视角的最小剂量。 **用户允许额外说明时，可在重拍单之外提供升级入口**： …
Read the whole file at its exact version.
How to install
mdr add flashfrogluo/second-take/second-take@sha256:438091c4ea355489The line at the top of this page pins the label v1.7.4, 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_wrvfbyggsvpyrxpq)
1 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 9 versions of this file, between Oct 12, 2026 and Oct 26, 2026, about one every 2 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| v1.7.4 latest | 2026-10-26 | 40cd639 | 52,626 B | A | view · diff |
| v1.7.3 | 2026-10-26 | 416770d | 52,626 B | A | view · diff |
| v1.7.3 | 2026-10-26 | 504f0c0 | 52,208 B | A | view · diff |
| v1.7.3 | 2026-10-26 | 79106d0 | 51,429 B | A | view · diff |
| v1.7.3 | 2026-10-18 | e6d285b | 50,913 B | A | view · diff |
| v1.7.3 | 2026-10-18 | f37e1bb | 49,978 B | A | view · diff |
| v1.7.3 | 2026-10-18 | 4630202 | 50,048 B | A | view · diff |
| v1.7.3 | 2026-10-18 | b387e0f | 49,937 B | A | view · diff |
| v1.7.2 | 2026-10-12 | 2416256 | 49,937 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 (52626 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
flashfrogluo/second-take · 3 stars · license MIT · pushed 2026-10-26 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_wrvfbyggsvpyrxpq GET https://markdownregistry.com/api/v1/resolve?ref=flashfrogluo/second-take/second-take GET https://markdownregistry.com/api/v1/blob/438091c4ea355489c164f6aed0c730c3572c61546c4ab27315046fb0660525cc
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.