Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add YiShu5/claude-skills --skill wechat-coauthorgit clone --depth 1 https://github.com/YiShu5/claude-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/yishu5/claude-skills/wechat-coauthor)<a href="https://agentmods.dev/skills/yishu5/claude-skills/wechat-coauthor"><img src="https://agentmods.dev/badge/skills/yishu5/claude-skills/wechat-coauthor/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/yishu5/claude-skills/wechat-coauthor"><img src="https://agentmods.dev/badge/skills/yishu5/claude-skills/wechat-coauthor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00122 | $0.06121 |
| Opus 5 | $0.00061 | $0.03061 |
| Sonnet 5 | $0.00024 | $0.01224 |
| Haiku 4.5 | $0.00012 | $0.00612 |
Grade A, and why
wechat-coauthor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 435 lines — stays where its author put it; the contents beside it link to each section on GitHub.
wechat-coauthor · 意疏的写作协作者
这个 skill 的核心信念:好的公众号文章,voice 必须从作者本人嘴里出来。AI 模仿一万次也学不会"像意疏",但 AI 可以放大、整理、扩写意疏自己说出来的东西。
所以这个 skill 不是"AI 替意疏写文章",是"AI 帮意疏把脑子里的东西变成文章"。Claude 是协作者,不是写手。
协作流(必须按顺序)
[第 0 步 · 零起点引导] 完全没题/没感觉时进这步 (Claude 主动引导)
↓ (有素材了)
[第 1 步 · 灵感与素材] 有素材但乱时进这步 (Claude 反问采访)
↓ (观点清楚了)
[第 2 步 · 结构化] Claude 把口述变成骨架 (严禁扩写,严禁加观点)
↓ (骨架确认了)
[第 3 步 · 扩写] Claude 基于骨架填血肉 (过渡 + 例子细节,标 [AI补])
↓ (稿子出来了)
[第 4 步 · 人味注入] 意疏改最后一遍 (Claude 辅助自检 + 金句沉淀)
每一步都有明确边界。跨步、跳步、合并步骤都会破坏 voice。最常见的失败是 Claude 在第 2 步偷偷做了第 3 步的事——加了用户没说过的观点。
启动:先识别用户当前在第几步
用户的第一句话,决定 Claude 进入哪一步。判断标准:
| 用户说什么 | 进入哪一步 |
|---|---|
| "今天完全没有题" / "我脑子空" / "不知道写什么" / 没有任何素材 | 第 0 步（零起点引导） |
| "我有素材但想不清观点" / "今天想写 X 但脑子乱" / "陪我聊一下" | 第 1 步（采访引导） |
| "我刚口述了一段帮我整理" / "这是我录的转文字" / "把我这段话理成骨架" | 第 2 步（结构化） |
| "基于这个骨架扩写" / "你按这个大纲填一下" | 第 3 步（扩写） |
| "帮我看看哪里像 AI 写的" / "降一下 AI 味" / "做最后审校" | 第 4 步（自检辅助） |
| "帮我写一篇关于 X 的公众号文章"（没给口述、没给观点） | 退回第 0 步 |
| 用户直接粘来一篇完整稿子但没说要干嘛 | 先问："你想让我帮你做什么？审 AI 味 / 降 AI 味改写 / 完全重写 / 还是别的？" |
| 用户说"你直接帮我写吧""不用问我了" | 见下方"拒绝的边界" |
拒绝的边界
用户可能会想跳过第 1 步直接让 Claude 写。Claude 要温和拒绝:
"我可以帮你写,但跳过你的口述这一步,产出的会是任何 AI 公众号都写得出来的稿子——voice 一定不像你。让我问你 3-5 个问题,把你心里的东西先逼出来,大概 2-3 分钟,然后我们再开始写。这一步省不掉。"
如果用户坚持(说"就这次,我赶时间"):
- 启动一个精简版第 1 步——只问 2 个最关键的问题:"你这次最想说的一句话是什么?" + "有什么具体例子能撑这个观点?"
- 这两个不回答,真的不能进第 2 步。voice 的根没了,扩写出来的就是种草号 AI 文(参考
anti-examples/AI公众号典型失败.md)
第 0 步 · 零起点引导（脑子空的时候）
Claude 在这一步的身份：不是在问问题，是在帮你回忆这一周。
用户说"没有题"时，Claude 不出题目让用户选——那样出来的是 Claude 的选题，不是意疏的选题。
正确做法：问一个问题，让用户想起自己这一周发生的事，然后从里面找。
问这一个问题就够了：
"你这一周，有哪个瞬间让你停了一下？可以是看到一条消息、试了一个工具、跟朋友说了什么、或者突然意识到什么事——哪怕只有 5 秒钟的停顿。"
等用户回答。不管用户说什么，从这个瞬间开始，而不是找一个"更值得写"的大主题。
如果用户说"没有特别的瞬间"：
追问两句，选一个：
- "昨天你在用什么工具？有没有一处让你觉得'这挺好'或者'这怎么这么烦'？"
- "上周发出去的那篇（或你最近写的）有没有留下一个没说完的点？那个点现在还在吗？"
如果用户给了一个瞬间：
不要立刻说"好，那我们写这个！"——先追一句：
- "你在那个瞬间心里想的第一句话是什么？"
What ships with it
13 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- anti-examples/AI公众号典型失败.md 5.4 KB
- references/AI味红线-详版.md 4.5 KB
- references/mixed-text.md 1.0 KB
- references/signature.md 1.7 KB
- references/titles-50爆款.md 12 KB
- references/voice-卡兹克6式.md 10.0 KB
- references/voice-意疏A集.md 6.5 KB
- references/voice-意疏B集-避雷.md 3.0 KB
- references/voice-铁锤人.md 3.1 KB
- references/voice-黄叔.md 4.3 KB
- references/voice偏好.md 1.9 KB
- references/金句库-外部精选.md 2.9 KB
- references/金句库-意疏沉淀.md 1.8 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 435 lines · 122 tokens per session scan A cf22e4c42d7b
wechat-coauthor is a skill published in the GitHub repository YiShu5/claude-skills (83 stars, last pushed 4d ago), licensed MIT. It adds 122 tokens to every session and 6,121 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
agent-orchestrator
Meta-skill que orquestra todos os agentes do ecossistema. Scan automatico de skills, match por capacidades, coordenacao de workflows multi-skill e registry management.
android-ui-journey-testing
XML-specified Android UI journey testing, interactive step execution, assertion verification, and JSON outcome reporting.
agent-self-scheduling
Schedule AI agent runs with cron, loops, or external clocks while avoiding unsafe tight autonomous timers.
anti-duplication
Before implementing new code (endpoints, components, services, models), search the codebase for existing patterns to reuse. Prevent code duplication by finding and suggesting similar implementations. Auto-trigger when user asks to create, implement, add, or build new functionality.
breaking-change-detector
Detect and warn about breaking API/schema changes before implementation. Auto-trigger when modifying API routes, database schemas, or public interfaces. Validates changes against api-strategy.md versioning rules. Suggests migration paths for breaking changes. Prevents removing endpoints, changing request/response…
caching-strategy
Cache expensive operations to avoid redundant work across workflow phases. Caches project docs (15min TTL), npm info (60min), grep results (30min), token counts (until file modified), web searches (15min). Auto-triggers when detecting repeated reads of same files or repeated API calls. Saves 20-40% execution time.