agent-starter-kit
llms.txt1 starsChanged 31 days agoApache-2.0View on GitHub
把仓库地址粘给你的 AI 助手(Claude Code / Cursor / Trae / Trae CN / Reasonix / Codex),它就自动装好 17 个技能、注册 MCP 工具、装 Python 依赖 —— 面向大学生与办公人群。 本文件是给 AI 的*只读地图*:告诉你该去读哪些文件。 行为约束与执行入口在 AGENTS.md,具体步骤在 INSTALL.md。
- Reads credentials
In this repository18 other files it gives its agents.
- AGENTS.mdAGENTS.md
- academic-figuresSkill
- citation-managementSkill
- cn-literatureSkill
- cpp-workflowSkill
- data-analysisSkill
- design-materialSkill
- doc-batchSkill
- email-calendarSkill
- lab-reportSkill
- math-modelingSkill
- meeting-notesSkill
- office-docsSkill
- paper-researchSkill
- python-workflowSkill
- study-tutorSkill
- thesis-formatSkill
- web-researchSkill
The file
Files
# agent-starter-kit > 把仓库地址粘给你的 AI 助手(Claude Code / Cursor / Trae / Trae CN / Reasonix / Codex),它就自动装好 17 个技能、注册 MCP 工具、装 Python 依赖 —— 面向大学生与办公人群。 本文件是给 AI 的**只读地图**:告诉你该去读哪些文件。 行为约束与执行入口在 [AGENTS.md](AGENTS.md),具体步骤在 [INSTALL.md](INSTALL.md)。 ## Docs - [AGENTS.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/AGENTS.md): **如果用户把本仓库地址发给你,先读这个。** 行为约束、允许自行判断的边界、禁止事项、特殊情况处理 - [INSTALL.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/INSTALL.md): 执行手册。Step 0 前置检查 → Step 1 环境探测 → Step 2 定能力包 → Step 3 打印清单等确认 → Step 4 安装 → Step 5 自检 → Step 6 配置用户项目的 AGENTS.md - [requirements.txt](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/requirements.txt): 安装引擎自身的依赖(`tomli-w` 用于写 TOML 宿主配置;`tomli` 用于 Python < 3.11 时读 TOML)。**建议先装**,否则 Reasonix / Codex 的 MCP 只能降级为手动粘贴 - [README.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/README.md): 给人看的门面。能力清单、支持的宿主、快速开始、FAQ - [uninstall.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/uninstall.md): 卸载指引。只删账本里记录过的条目,不碰用户自己的配置 - [SECURITY.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/SECURITY.md): 安全披露政策与本项目的风险边界 - [presets.json](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/presets.json): 场景预设 —— 拿用户原话去匹配 `trigger`,决定装哪些能力包 - [hosts.json](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/hosts.json): 宿主适配表。技能目录、MCP 配置路径、写入策略;含未核实路径的标注 - [packs/SCHEMA.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/packs/SCHEMA.md): 能力包的数据格式。字段名写错会被静默忽略 - [community-skills.json](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/community-skills.json): 社区高星技能索引(带 star 与许可证风险)。默认不装 - [CHANGELOG.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/CHANGELOG.md): 版本变更记录 ## Optional - [templates/AGENTS.md.template](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/templates/AGENTS.md.template): 给**用户自己的项目**生成 AGENTS.md 时用的模板(60 行以内) - [templates/claude-settings.json.template](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/templates/claude-settings.json.template): Claude Code 权限白名单模板(allow 安全命令 + deny 屏蔽 .env) - [templates/README.md](https://raw.githubusercontent.com/tyiiiyt01/agent-starter-kit/main/templates/README.md): 上面两个模板的用法说明 - [packs/](https://github.com/tyiiiyt01/agent-starter-kit/tree/main/packs): 八个能力包定义(code / research / math-modeling / writing / office / office-work / study / design) - [scripts/](https://github.com/tyiiiyt01/agent-starter-kit/tree/main/scripts): detect.py(只读探测)· apply.py(幂等安装)· doctor.py(逐项自检) ## 关键约束(摘要,完整版见 AGENTS.md) - **唯一硬前置是 Python 3**。缺了就停下来引导用户安装,不要继续 - 安装前必须先 `--dry-run` 打印清单,**等用户明确确认**才动手 - **不写系统级配置,不用管理员权限**,一律装到用户目录 - **不改 `skills/` 里技能文件的内容** —— 必须逐字节一致 - **不要主动加装 MCP** —— MCP 极吃上下文(实测 3 个可占 72% 的 200k 上下文) - 大件依赖(LibreOffice 等 heavy 档)**永不自动安装** - 装完由 `doctor.py` 判定成败,**不要跳过**
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

