Skip to content

模块化 AI 编程技能系统,将混乱的智能体输出转化为结构化、可追溯、生产就绪的代码 「智能路由 + 工作流 + 知识库 + 命令驱动」 不仅仅是编程!

License

Notifications You must be signed in to change notification settings

nightwhite/helloagents

 
 

Repository files navigation

HelloAGENTS v2 (helloagents_v2 skill)

This repo ships one unified dev/maintenance skill: helloagents_v2/. It supports auto routing + end-to-end workflow + per-project command permissions via helloagents/config.toml.

中文:README_CN.md


Install / Deploy (manual copy)

1) Get the repo

git clone https://github.com/nightwhite/helloagents.git
cd helloagents

2) Copy the skill into your AI CLI config directory

Copy helloagents_v2/ to:

Source Target
helloagents_v2/ <CLI_CONFIG_ROOT>/skills/helloagents_v2/

Common <CLI_CONFIG_ROOT>:

  • Claude Code: ~/.claude/
  • Codex CLI: ~/.codex/

macOS/Linux (Claude Code):

cp -R helloagents_v2 ~/.claude/skills/helloagents_v2

macOS/Linux (Codex CLI):

cp -R helloagents_v2 ~/.codex/skills/helloagents_v2

Windows PowerShell (Claude Code example):

Copy-Item -Recurse helloagents_v2 $env:USERPROFILE\.claude\skills\helloagents_v2

3) Restart & verify

Restart your AI CLI, then enter/trigger the helloagents_v2 skill first (how you do this depends on your CLI: pick the skill, or explicitly invoke the skill name in chat).

After explicitly invoking the helloagents_v2 skill, run:

~help

Per-project permissions (recommended)

Each project can allow/deny commands via:

  • Path: <PROJECT_ROOT>/helloagents/config.toml
  • PROJECT_ROOT defaults to your CLI working directory (cwd)

If this file is missing, the skill will auto-initialize it before running any command (default: allow all commands).

Manual initialization (optional):

mkdir -p helloagents
cp <CLI_CONFIG_ROOT>/skills/helloagents_v2/assets/templates/config.toml helloagents/config.toml

Modes:

  • denylist: allow everything by default, disable risky commands in deny = ["~rollback", ...]
  • allowlist: only allow commands listed in allow = ["~plan", "~review"]

Common commands (minimal)

  • ~help: show allowed commands for this project + descriptions
  • ~plan <request>: run until design and generate a package (stops before develop)
  • ~exec: pick and run a package under helloagents/plan/ (handles unarchived/incomplete)
  • ~auto <request>: run end-to-end (subject to rules & permissions)
  • ~review / ~test / ~validate: review / test / validate
  • ~rollback: rollback (recommended to disable per project)

Upgrade / Uninstall

Upgrade:

  • Replace <CLI_CONFIG_ROOT>/skills/helloagents_v2/ with the new version
  • Your project helloagents/config.toml stays intact

Uninstall:

  • Delete <CLI_CONFIG_ROOT>/skills/helloagents_v2/

About

模块化 AI 编程技能系统,将混乱的智能体输出转化为结构化、可追溯、生产就绪的代码 「智能路由 + 工作流 + 知识库 + 命令驱动」 不仅仅是编程!

Resources

License

Contributing

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 100.0%