nagi-studio
Nagi Studio's agent skills, as an installable Claude Code plugin.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/nagi-studio--skills)HTML
<a href="/listings/nagi-studio--skills"><img src="/api/listings/nagi-studio--skills/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
把这句话发给你的 agent:
帮我安装 https://github.com/nagi-studio/skills 里的 skills
> **Agent 请照做:** 运行 `npx skills@latest add nagi-studio/skills -g -y` 安装到用户全局目录。如果你是 Claude Code,也可以让用户运行 `/plugin marketplace add nagi-studio/skills` 和 `/plugin install nagi-skills@nagi-studio`。装完列出已安装的 skill,并提醒用户重启会话。
自己装:
npx skills@latest add nagi-studio/skills -g
用的是 [skills CLI](https://github.com/vercel-labs/skills),会自动识别本机装了哪些 agent(Claude Code、Codex、Cursor、Antigravity 等),装进各自的 skills 目录。`-g` 表示装到全局,所有项目都能用;去掉它就只装进当前项目。加 `-a <agent>` 可以指定某个 agent。
<details> <summary>Claude Code 也可以用插件装</summary>
/plugin marketplace add nagi-studio/skills /plugin install nagi-skills@nagi-studio
</details>
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 17 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- {
- "name": "nagi-studio",
- "owner": {
- "name": "Nagi Studio",
- "url": "https://github.com/nagi-studio"
- },
- "description": "Nagi Studio's agent skills, as an installable Claude Code plugin.",
- "plugins": [
- {
- "name": "nagi-skills",
- "source": "./",
- "description": "Companion skills for the Bilibili series Re:从零开始的AI学习. Each skill has an episode that explains why.",
- "category": "productivity",
- "keywords": ["skills", "agent-skills", "claude-code", "codex"]
- }
- ]
- }
README.md · README · 92 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- # Nagi Skills
- **中文** · [English](./README.en.md)
- 配套《Re:从零开始的AI学习》,每个 skill 都有一期视频讲为什么这么做。[📺 合集](https://space.bilibili.com/312249633/lists/8719249?type=season)
- ## 安装
- 把这句话发给你的 agent:
- ```
- 帮我安装 https://github.com/nagi-studio/skills 里的 skills
- ```
- > **Agent 请照做:** 运行 `npx skills@latest add nagi-studio/skills -g -y` 安装到用户全局目录。如果你是 Claude Code,也可以让用户运行 `/plugin marketplace add nagi-studio/skills` 和 `/plugin install nagi-skills@nagi-studio`。装完列出已安装的 skill,并提醒用户重启会话。
- 自己装:
- ```bash
- npx skills@latest add nagi-studio/skills -g
- ```
- 用的是 [skills CLI](https://github.com/vercel-labs/skills),会自动识别本机装了哪些 agent(Claude Code、Codex、Cursor、Antigravity 等),装进各自的 skills 目录。`-g` 表示装到全局,所有项目都能用;去掉它就只装进当前项目。加 `-a <agent>` 可以指定某个 agent。
- <details>
- <summary>Claude Code 也可以用插件装</summary>
- ```bash
- /plugin marketplace add nagi-studio/skills
- /plugin install nagi-skills@nagi-studio
- ```
- </details>
- ## 更新
- 装好的 skill 不会自己跟着仓库更新。仓库有新版本时,把这句话发给你的 agent:
- ```
- 帮我更新 https://github.com/nagi-studio/skills 里的 skills
- ```
- > **Agent 请照做:** 对这个仓库里已安装的每个 skill 运行 `npx skills@latest update <skill 名> -g -y`。如果用户是用 Claude Code 插件装的,运行 `claude plugin marketplace update nagi-studio` 和 `claude plugin update nagi-skills@nagi-studio`。更新完提醒用户重启会话。
- 自己更新:
- ```bash
- npx skills@latest update grill -g
- ```
- `grill` 换成要更新的 skill 名;不写名字会更新本机所有全局 skill,不只是这个仓库的。
- <details>
- <summary>用 Claude Code 插件装的</summary>
- 在终端里运行:
- ```bash
- claude plugin marketplace update nagi-studio
- claude plugin update nagi-skills@nagi-studio
- ```
- 已经开着的会话运行 `/reload-plugins`,或者重启。
- </details>
- ## Skills
- ### Pondering
- | Skill | 做什么 | 视频 |
- |---|---|---|
- | [`grill`](./skills/pondering/grill) | 分轮追问你的计划或想法,直到重要未知暴露出来。5 档强度,任何领域都能用。<br><sub>来源:思路受 Matt Pocock 的 [grilling](https://github.com/mattpocock/skills/tree/main/skills/productivity/grilling)(MIT)启发,重写为面向所有人的通用版:中文、5 档追问强度、问到「剩下的未知不再影响下一步」就停。</sub> | [<img src="./assets/videos/BV1irHx6dEEf.jpg" width="240" alt="📺">](https://www.bilibili.com/video/BV1irHx6dEEf) |
- ### Rookie
- 适合刚开始用 agent 做东西的新手。这几个其实是练脑子的:装不装 skill 不重要,重要的是把这套想法记在脑子里——先比方案再动手、让 agent 拿出证据而不是只信它一句「做好了」、交差前对着清单做减法、项目大了就给它大扫除。记住了,换什么 agent、装不装这几个 skill,你都会这样要求它。
- | Skill | 做什么 | 视频 |
- |---|---|---|
- | [`architect`](./skills/rookie/architect) | 需求清楚之后,比较两三套技术方案并推荐一套。按你管得过来的程度排序,每套写明你得自己管哪些密钥和权限。接着 grill 的决策记录往下做。 | [<img src="./assets/videos/BV1TtpL6SEvg.jpg" width="240" alt="📺">](https://www.bilibili.com/video/BV1TtpL6SEvg) |
- | [`prove-it`](./skills/rookie/prove-it) | 让 agent 拿证据交差,而不是一句「做好了」:交差前跑测试、在真浏览器里点一遍并截图;同一个错被纠正第二次,就把它变成下次一犯就报错的检查;顺手清理三种垃圾测试。<br><sub>来源:「把错误变成会报错的检查」这个思路来自 Poteto 的 [correct](https://github.com/cursor/plugins/tree/main/pstack/skills/correct);三种垃圾测试的判断参考了 Matt Pocock 的 [tdd](https://github.com/mattpocock/skills/tree/main/skills/engineering/tdd);浏览器里点一遍再截图的做法,来自 [Voyager](https://github.com/voyager-crew/voyager) 专用的 [verify-in-browser](https://github.com/voyager-crew/voyager/blob/main/.agents/skills/verify-in-browser/SKILL.md),改写成了任何网页项目都能用的通用版。</sub> | [<img src="./assets/videos/BV1BgH36DE7y.jpg" width="240" alt="📺">](https://www.bilibili.com/video/BV1BgH36DE7y) |
- | [`de-slop`](./skills/rookie/de-slop) | 给做好的界面去 AI 味:截图,对着清单一条条做减法,改到每条都过;清单参考了 [@dingyi](https://x.com/dingyi/status/2115790005423734818) 整理的 AI 味前端特征。 | [<img src="./assets/videos/BV1pbxr6REJi.jpg" width="240" alt="📺">](https://www.bilibili.com/video/BV1pbxr6REJi) |
- | [`tidy`](./skills/rookie/tidy) | 项目变大、agent 开始改东忘西时,给项目做一次大扫除:同一条规矩收成一处,清掉死代码和废话注释,装上超过 5,000 token 就拒绝提交的钩子,给 AGENTS.md 和工具瘦身。<br><sub>来源:文件大小钩子来自 Matt Pocock 的[提交前钩子](https://x.com/mattpocockuk/status/2111708560715292675)。</sub> | [<img src="./assets/videos/BV1Gu4L6iEdc.jpg" width="240" alt="📺">](https://www.bilibili.com/video/BV1Gu4L6iEdc) |
- ## 维护
- ```bash
- ./scripts/link-skills.sh
- ```
- 把仓库里的 skill 直接软链到各 agent 的全局 skills 目录,之后在仓库里 `git pull` 即可更新,不经过上面的更新命令。新增 skill 的约定见 [AGENTS.md](./AGENTS.md)。
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.