skill-refiner SKILL.md skillA
skill-refiner from functoreality/agent-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add functoreality/agent-skills/skill-refiner@v260804Pins v260804 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- functoreality/agent-skills
- Path
skills/skill-refiner/SKILL.md- Versions recorded
- 6
- Latest version
- v260804, committed Oct 29, 2026
- First version recorded
- Aug 2, 2026
- Size
- 26,723 bytes
- Repository stars
- 11
- License
- LGPL-3.0
通过复盘反思或专家输出对比，发现并提炼思维/写作/设计型 skill 的改进方向。 适用于分析问题、写高质量笔记等产出质量无客观指标，需要靠推演和专家判断来评估的 skill。 当用户要求审查、改进、更新或复盘某个 skill，或提供了专家输出用于与 AI 输出对比，或对某次 skill 执行的质量给出了具体反馈时，使用本 skill。.
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
# Skill Refiner ## 概述 本 skill 用于改进**思维/写作/设计型 skill**——那些产出质量无法用客观指标衡量、需要靠推演和专家判断来评估的 skill。 不适用于可用指标或成功/失败信号自动评估的执行型 skill（部署、代码生成等）。 核心命题：当"什么是好"本身也需要逐步发现时，如何系统化地改进一个 skill？ ### 目标 skill 的特征 - 产出是文本文档或分析结果，质量是主观的、审美的 - 没有标准答案或客观评估指标 - 判断产出好坏需要推演使用场景（"用户读到这份产出后会怎么想？能否受益？"） - 很多时候，改进方向不是修正"错误"，而是发现"还可以有更好的分析方式" ### 辅助文件 - `references/quality-lenses.md`：质量维度透镜的详细说明。在归因阶段需要从不同角度审视差异时查阅。 - 目标 skill 目录下的 `learnings.md`：该 skill 的工作记忆，记录尚未处理的原始观察。仅在阶段四出现可升级类弱观测时按 §4.4 交叉验证读取，不在复盘开始时读取。见下文"learnings.md 机制"。 --- ## 三种工作模式 ### 模式 A：专家对比分析 用户同时提供了 AI 输出和专家输出（或用户自己修改后的版本）。 不只比较输出形式的差异，而是追问：专家在写出这些内容时，脑中经过了什么样的思维过程，而 AI 没有？ 差异可能体现在多个方面，但不限于以下角度——你看到的任何差异都值得探究： - **预判差异**：专家做了 AI 没做的预判（如预判读者会在什么场景下查阅这份产出） - **取舍差异**：专家在某个节点做了 AI 没做的取舍（如选择深入 A 而放弃 B，背后的考量是什么） - **框架差异**：专家用了 AI 没用到的组织框架（如换了一个概念维度来归类信息） - **你发现的任何其他差异**，只要追溯到思维层面的区别，都可能指向 skill 的改进方向 **差异分析的粒度要求**：在进行专家对比时，不能仅按行比较。 一行的修改中可能包含多个独立改动（换词、调序、增删限定语、拆句、调整层级结构，等等）， 按行分析会将其笼统归为一个差异点，遗漏大量信息。 必须将差异拆解为**原子改动**——每个原子改动对应文本的一个独立变化维度，无法再分解为更小的独立操作。 在此基础上对每个原子改动分别归因。 ### 模式 B：执行复盘反思 AI 刚用某个 skill 完成了一次任务，用户要求复盘。 回顾执行过程，寻找任何值得改进的信号。 以下是一些常见的入手角度，但不要局限于此——任何你觉得可能意味着 skill 应该调整的地方都值得提出： …
Read the whole file at its exact version.
How to install
mdr add functoreality/agent-skills/skill-refiner@sha256:bcb9691ade90740aThe line at the top of this page pins the label v260804, 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_cfoofb74k23yggu4)
2 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 6 versions of this file, between Aug 2, 2026 and Oct 29, 2026, about one every 18 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| v260804 latest | 2026-10-29 | 08737da | 26,723 B | A | view · diff |
| v260804 | 2026-08-05 | b5e2e41 | 25,836 B | A | view · diff |
| git:20260804.d06f215 | 2026-08-04 | d06f215 | 25,774 B | A | view · diff |
| git:20260802.f657aa5 | 2026-08-02 | f657aa5 | 23,217 B | A | view · diff |
| git:20260802.ba3e509 | 2026-08-02 | ba3e509 | 21,610 B | A | view · diff |
| git:20260802.b6b7507 | 2026-08-02 | b6b7507 | 21,751 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 (26723 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
functoreality/agent-skills · 11 stars · license LGPL-3.0 · pushed 2026-11-05 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_cfoofb74k23yggu4 GET https://markdownregistry.com/api/v1/resolve?ref=functoreality/agent-skills/skill-refiner GET https://markdownregistry.com/api/v1/blob/bcb9691ade90740a14fc81bf28c43c37582846e1148b7828727fca95eb29bc61
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.
More from functoreality/agent-skills
Every file in functoreality/agent-skills