菜单
Install or update the skills toolkit (skills, rules, templates, prompts, CLAUDE.md) in the current project from the canonical GitHub repo. Use when starting a new project, when asked to "install skills", "set up the toolkit", "update skills to latest", or "bootstrap".
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/matheuspoleza/skills --skill bootstrap命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Implement a milestone autonomously inside a Conductor worktree without stop-and-ask checkpoints, strictly enforcing scope-first and logging anything out of scope. Use when running parallel Conductor lanes, asked to "build autonomously", "run this in the background", "yolo this milestone", or for low-risk batched UX/tech-debt work.
Implement a milestone's tasks one at a time using BDD then TDD, with live progress tracked in three places and a Review Contract filled per task. Use when ready to implement, given a milestone path, or asked to "build the next milestone" or "start implementing".
Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.
基于 SOC 职业分类
正在显示 SKILL.md
| name | bootstrap |
| description | Install or update the skills toolkit (skills, rules, templates, prompts, CLAUDE.md) in the current project from the canonical GitHub repo. Use when starting a new project, when asked to "install skills", "set up the toolkit", "update skills to latest", or "bootstrap". |
Fetch the latest skills toolkit and install it into the current project. Works for first install and as the update command for existing installs.
In the target project root:
.claude/skills/, .claude/rules/, .claude/templates/, .claude/prompts/ populated with the latest contentsCLAUDE.md created, or rule references appended to an existing filedocs/ directory ensuredpackage.json is detected (devDependency + .mcp.json entry)CLAUDE.md (appended to, not overwritten)package.json under the target (to detect React for Agentation).mcp.json at the project root (created or updated)No project source code is read or modified.
Default target is the current working directory. If the user names a different target, take its absolute path. Confirm before running if the target isn't obvious.
From the target directory:
curl -fsSL https://raw.githubusercontent.com/matheuspoleza/skills/main/install.sh | bash
The script self-bootstraps: it clones the repo into a tempdir and re-execs from there. Requires git and curl (both standard).
For a non-CWD target, pass the path explicitly:
curl -fsSL https://raw.githubusercontent.com/matheuspoleza/skills/main/install.sh | bash -s -- /path/to/project
Summarize what was installed or updated from the installer's stdout — skills added, rules appended to CLAUDE.md, deprecated entries removed. Surface any warnings verbatim, especially:
npm install agentation -D failed — user must run it manuallyfailed to update .mcp.json — likely malformed existing JSON/onboard, /research, etc.) — the user invokes those after install