测试助手

Name: 测试助手
Author: tomevault-io

AI 智能社区

作者仓库星标 0
作者更新于实时读取
作者仓库 skills-registry

流狐档案事实层信息：领域、Agent、信任分、运行环境、权限与网络

领域: AI 智能
兼容 Agent: Claude Code
Cursor
Cline
Codex
Windsurf
Gemini CLI

+20
信任分: 83 / 100 · 社区维护
作者 / 版本 / 许可: @tomevault-io · 未声明 license
Token 消耗评级: 低消耗
接入复杂程度: 需手动接入
是否需要外部 API Key: 需要 · OpenAI / Anthropic / Google / Gemini
兼容的系统: macOS
底层运行要求: 无特殊要求
文件与系统权限: 只读
允许写入 / 修改
Shell 执行
读取环境变量
网络行为: 允许外网请求
安装命令数: 26 条

档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。

需要注意： 未限定 allowed-tools，默认拥有全部工具权限。；检出高风险片段：pipe_curl_to_shell

输出预览 cli-ai-tools.preview

---
name: cli-ai-tools
description: This skill should be used when the user asks about "CLI AI tools", "Claude Code", "Codex CLI", "…
category: AI 智能
runtime: 无特殊运行时
---

# cli-ai-tools 输出预览

## PART A: 任务判断
- 适用问题：提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求：目标材料、限制条件、期望输出和验收方式。
- 证据边界：围绕“Tool Index / Anthropic: Claude Code / OpenAI: Codex CLI”读取原文规则，不把推断写成作者承诺。

## PART B: 执行结果
- **01** 任务判断：确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理，并标出输入、限制和预期结果。
- **02** 执行计划：优先按“Tool Index / Anthropic: Claude Code / OpenAI: Codex CLI”拆成步骤，说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果：给出可复制的命令、文件改动、检查清单或内容草稿，并说明如何继续迭代。
- **04** 风险边界：结合 读取文件、写入/修改文件、执行终端命令、读取环境变量、会按任务需要访问外部网络、需要准备 OpenAI / Anthropic / Google / Gemini API Key 给出执行前确认项。

## Running Rules
- 读取文件、写入/修改文件、执行终端命令、读取环境变量；会按任务需要访问外部网络；需要准备 OpenAI / Anthropic / Google / Gemini API Key。
- 先小样例验证，再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。

SKILL.md SKILL.md

---
name: cli-ai-tools
description: This skill should be used when the user asks about "CLI AI tools", "Claude Code", "Codex CLI", "…
category: AI 智能
source: tomevault-io/skills-registry
---

# cli-ai-tools

## 什么时候使用
- 把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。 把任务拆成可执行、可检查…
- 面向提示词、Agent 工作流、模型评估或自动化推理，优先处理能明确输入、步骤和验收标准的工作。

## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。

## 执行规则
- 围绕「Tool Index / Anthropic: Claude Code / OpenAI: Codex CLI」组织步骤，不把推断写成作者事实。
- 读取文件、写入/修改文件、执行终端命令、读取环境变量；会按任务需要访问外部网络；需要准备 OpenAI / Anthropic / Google / Gemini API Key。
- 先跑小样例，确认结果可检查后再扩大任务范围。

## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown，不编造命令、平台或依赖。

证据边界与执行链路

作者原文负责流程事实；仓库文件负责来源和命令；流狐只补充适用场景、限制和质量判断。

skill "cli-ai-tools" {
  输入层     -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
  上下文层   -> Tool Index / Anthropic: Claude Code / OpenAI: Codex CLI
  规则层     -> SKILL.md 触发条件 / 执行顺序 / 输出格式
  运行层     -> 无特殊运行时 | 读取文件、写入/修改文件、执行终端命令、读取环境变量 | 会按任务需要访问外部网络
  安全层     -> 需要准备 OpenAI / Anthropic / Google / Gemini API Key + 小任务验证 + diff / 日志复核
  输出层     -> 可复制结果 + 检查清单 + 下一步迭代
}

解读按原文结构重写，命令、链接、术语均保留；右侧可核对作者原始 SKILL.md

先判断是否适合

核心用途：把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。把任务拆成可执行、可检查、可继续迭代的步骤…
适用任务：更适合 AI Agent、提示词、模型评估与自动化推理，尤其是输入、步骤、验收口径可以反复复用的场景。
不宜硬套：如果任务只有一次性问答、来源材料很少，或需要作者没有说明的命令和外部服务，应先降级为普通对话。

作者设计意图

触发依据：使用时机、上下文收集、输出格式和失败处理共同决定它是否适合当前任务。
来源边界：作者原文负责事实，仓库文件负责命令和文件证据；Fluxly 只补充用途、限制和使用判断。

作者的方法与取舍

执行主线：任务围绕「Tool Index / Anthropic: Claude Code / OpenAI: Codex CLI」组织，把触发条件、必要上下文、执行规则和交付标准连成一条可复核的链路。
设计重点：重点不在重复教程，而在解释作者为什么规定这些步骤、哪些条件不可省略，以及结果偏离时应回到哪一层检查。
使用判断：先确认任务与原文边界一致，再决定是否启用；来源没有说明的命令、平台和依赖保持未知，不由 Fluxly 补写。

边界和复核

依赖条件：需要提前准备 OpenAI / Anthropic / Google / Gemini API Key，未配置前不要直接跑完整任务。
权限提醒：会涉及 read / write / shell-exec / env-read 权限，执行前要说明可读写范围、命令边界和回退方式。
质量标准：结果应包含明确产物、必要证据和下一步动作；如果只是泛泛解释，需要补充输入和验收条件后重跑。

CLI AI Coding Tools Reference

Comprehensive reference for all major command-line AI coding assistants. This covers installation, key commands, flags, configuration, and model selection for each tool.

Tool Index

Tool	Provider	Primary Use
Claude Code	Anthropic	Agentic coding, full codebase understanding
Codex CLI	OpenAI	Agentic coding with sandboxed execution
Gemini CLI	Google	Agentic coding with Gemini models
Aider	Open Source	AI pair programming, git-integrated
Goose	Block	Extensible AI agent with plugins
amp	Sourcegraph	AI agent for large codebases
Copilot CLI	GitHub	Command-line suggestions and explanations
Cody CLI	Sourcegraph	Codebase-aware AI assistant
OpenCode	Open Source	Go-based terminal AI coding
qodo	Qodo	AI code quality and testing
Jules	Google	Async coding agent (GitHub-integrated)
Continue	Open Source	AI code assistant (IDE + CLI)
avante	Open Source	Neovim AI plugin (CLI-adjacent)

Anthropic: Claude Code

The CLI agent currently running in this session. Agentic AI coding with deep codebase understanding, tool use, and MCP integration.

Install:

npm install -g @anthropic-ai/claude-code

Basic usage:

claude                          # interactive REPL
claude "explain this codebase"  # start with a prompt
claude -p "what does main do"   # print mode (non-interactive, outputs to stdout)
claude --resume                 # resume last conversation
claude --continue               # continue last conversation with new input

Key flags:

Flag	Description
`--model <model>`	Select model (e.g. `claude-sonnet-4-20250514`)
`-p, --print`	Print mode: non-interactive, single response to stdout
`--resume`	Resume the most recent conversation
`--continue`	Continue last conversation with new input
`--dangerously-skip-permissions`	Skip all permission prompts (use with caution)
`--allowedTools <tools>`	Comma-separated list of allowed tools
`--output-format <fmt>`	Output format: `text`, `json`, `stream-json`
`--max-turns <n>`	Limit autonomous turns in non-interactive mode
`--verbose`	Enable verbose logging

Slash commands (interactive mode):

Command	Description
`/help`	Show all available commands
`/compact`	Compress conversation to save context
`/clear`	Clear conversation history
`/config`	View/edit configuration
`/permissions`	Manage tool permissions
`/install-plugin <url>`	Install a plugin from GitHub
`/mcp`	Manage MCP server connections
`/model`	Switch model mid-conversation
`/cost`	Show token usage and cost
`/doctor`	Diagnose installation issues

Configuration:

Project config: .claude/settings.json in project root
User config: ~/.claude/settings.json
Memory: CLAUDE.md files (project root, ~/.claude/CLAUDE.md)
MCP servers: .claude/mcp.json

Environment variables:

ANTHROPIC_API_KEY=sk-ant-...        # API key
CLAUDE_MODEL=claude-sonnet-4-20250514  # Default model
ANTHROPIC_BASE_URL=...              # Custom API endpoint

OpenAI: Codex CLI

OpenAI's open-source agentic coding CLI. Runs code in a sandboxed environment with network-disabled containers.

Install:

npm install -g @openai/codex

Basic usage:

codex                           # interactive mode
codex "refactor the auth module"  # start with prompt
codex --approval-mode full-auto "fix all tests"  # fully autonomous

Key flags:

Flag	Description
`--model <model>`	Select model (default: `codex-mini-latest`)
`--approval-mode <mode>`	`suggest` (default), `auto-edit`, `full-auto`
`--quiet`	Minimal output
`-p, --prompt`	Non-interactive single prompt

Approval modes:

suggest — asks before every action
auto-edit — auto-approves file edits, asks for commands
full-auto — auto-approves everything (sandboxed)

Configuration:

Config file: ~/.codex/config.yaml or ~/.codex/config.json
Instructions: AGENTS.md in project root (similar to CLAUDE.md)

Environment variables:

OPENAI_API_KEY=sk-...               # API key
CODEX_MODEL=codex-mini-latest       # Default model

Google: Gemini CLI

Google's CLI coding assistant powered by Gemini models. Free tier available with Google account.

Install:

npm install -g @google/gemini-cli
# Or run without installing:
npx @google/gemini-cli

Basic usage:

gemini                          # interactive mode
gemini "explain this function"  # start with prompt
gemini -p "summarize README"   # non-interactive print mode

Key flags:

Flag	Description
`--model <model>`	Select model (e.g. `gemini-2.5-pro`)
`-p`	Non-interactive print mode
`--sandbox`	Run in sandboxed mode
`--debug`	Enable debug output

Configuration:

Config: ~/.gemini/settings.json
Memory: GEMINI.md in project root
Supports MCP servers via .gemini/mcp.json

Environment variables:

GOOGLE_API_KEY=...                  # API key (or use Google auth)
GEMINI_MODEL=gemini-2.5-pro        # Default model

Google: Jules

Google's asynchronous coding agent. Integrates with GitHub — tasks are assigned via issues and Jules works on them in the background, creating PRs.

Usage:

Accessed via jules.google.com or GitHub integration
Not a traditional CLI tool — operates asynchronously on GitHub repos
Assign tasks via GitHub issues or the Jules web interface
Jules creates branches, makes changes, submits PRs

Key features:

Async operation (runs independently in the background)
GitHub-native integration
Multi-file changes with PR creation
Powered by Gemini models

Open Source: Aider

The most popular open-source AI pair programming CLI. Deep git integration, multi-model support, architect/editor pattern.

Install:

pip install aider-chat
# Or:
pipx install aider-chat
# Or with uv:
uv tool install aider-chat

Basic usage:

aider                               # interactive, auto-detects git repo
aider --model claude-3.5-sonnet     # specify model
aider file1.py file2.py             # add files to context
aider --architect                   # architect mode (plan then edit)

Key slash commands (interactive):

Command	Description
`/add <file>`	Add file(s) to the chat context
`/drop <file>`	Remove file(s) from context
`/run <cmd>`	Run a shell command and share output
`/diff`	Show pending diffs
`/commit`	Commit changes with AI-generated message
`/undo`	Undo the last change
`/model <name>`	Switch model
`/architect`	Toggle architect mode
`/ask`	Ask without editing (question-only mode)
`/lint`	Lint changed files
`/test`	Run tests
`/tokens`	Show token usage
`/clear`	Clear chat history
`/exit`	Exit aider

Key flags:

Flag	Description
`--model <model>`	Main model for editing
`--architect`	Use architect mode (separate planning model)
`--editor-model <model>`	Model for code edits (in architect mode)
`--auto-commits` / `--no-auto-commits`	Toggle auto-commit after each change
`--watch-files`	Watch files for changes and auto-add
`--no-git`	Disable git integration
`--yes`	Auto-confirm prompts
`--edit-format <fmt>`	`whole`, `diff`, `udiff`, `diff-fenced`

Configuration:

Config file: .aider.conf.yml (project root or ~/.aider.conf.yml)
Environment: .env file in project root
Model aliases: .aider.model.settings.yml
Environment variables: ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY, OPENROUTER_API_KEY, AIDER_MODEL

Block: Goose

Block's open-source AI agent CLI. Extensible with a plugin system called "extensions" (formerly "toolkits").

Install:

# macOS
brew install block/tap/goose

# Or from source
cargo install goose-cli

# Or via installer script
curl -fsSL https://github.com/block/goose/releases/latest/download/install.sh | bash

Basic usage:

goose session             # start interactive session
goose session -r          # resume last session
goose run "fix tests"     # run with a prompt
goose configure           # configure settings

Key commands:

Command	Description
`goose session`	Start a new session
`goose session -r`	Resume last session
`goose run <prompt>`	Run a single task
`goose configure`	Interactive configuration
`goose info`	Show current configuration

Configuration:

Config file: ~/.config/goose/config.yaml
Supports multiple providers: OpenAI, Anthropic, Google, Ollama, local models
Extensions (plugins) for additional capabilities

Sourcegraph: amp

Sourcegraph's AI coding agent for the terminal. Designed for large codebase understanding.

Install:

# Download from https://ampcode.com
# Or via npm (check current package name):
npm install -g @anthropic-ai/amp

Basic usage:

amp                         # interactive mode
amp "explain the auth flow" # start with prompt
amp --continue              # continue last conversation

Key features:

Thread-based conversations (similar to Claude Code)
Codebase indexing and search
Multi-file editing
Git-aware operations

Configuration:

Config: ~/.amp/settings.json
Memory: AMP.md in project root

GitHub: Copilot CLI

GitHub Copilot's CLI integration. Provides command suggestions and explanations via the gh CLI extension.

Install:

gh extension install github/gh-copilot

Key commands:

Command	Description
`gh copilot suggest "find large files"`	Get command suggestions
`gh copilot explain "tar -xzf file.tar.gz"`	Explain a command

Flags:

Flag	Description
`-t shell`	Target: generic shell command
`-t git`	Target: git command
`-t gh`	Target: GitHub CLI command

Examples:

gh copilot suggest -t git "squash last 3 commits"
gh copilot explain "find . -name '*.log' -mtime +30 -delete"
gh copilot suggest -t shell "monitor CPU usage"

Sourcegraph: Cody CLI

Sourcegraph's AI coding assistant with CLI capabilities. Strong at codebase search and understanding.

Install:

npm install -g @sourcegraph/cody

Basic usage:

cody chat "explain this repo"    # ask questions
cody auth login                  # authenticate
cody auth status                 # check auth

Configuration:

Auth: Sourcegraph access token or SRC_ACCESS_TOKEN env var
Endpoint: SRC_ENDPOINT for self-hosted Sourcegraph

Open Source: OpenCode

Go-based AI coding assistant for the terminal. Lightweight TUI with multi-provider support.

Install:

go install github.com/opencode-ai/opencode@latest
# Or download binary from releases

Basic usage:

opencode                        # launch TUI

Configuration:

Config file: opencode.json in project root
Supports: OpenAI, Anthropic, Google, Groq, local models
Provider-specific API keys via environment variables

Qodo: qodo

AI-powered code quality and test generation tool.

Install:

pip install qodo
# Or:
npm install -g qodo

Key commands:

qodo test <file>                # generate tests
qodo review <file>              # code review
qodo improve <file>             # suggest improvements

Neovim: avante.nvim

AI-powered Neovim plugin. CLI-adjacent — operates within Neovim's terminal-based interface.

Install: Add to Neovim plugin manager (lazy.nvim, packer, etc.):

{ "yetone/avante.nvim", build = "make" }

Key bindings (in Neovim):

<leader>aa — Open Avante chat
<leader>ae — Edit selected code with AI
<leader>ar — Refresh AI response

Configuration:

In Neovim config (init.lua)
Supports: OpenAI, Anthropic, Azure, Ollama, Copilot

Other CLI-Adjacent Tools

The following are primarily IDE-based, not standalone CLI agents:

Tool	Type	Notes
Cline / Roo Code	VS Code extension	Cline forked as Roo Code. No standalone CLI binary.
Cursor	IDE (VS Code fork)	Open from terminal with `cursor` command.
Windsurf	IDE (Codeium)	Open from terminal with `windsurf` command.
Trae	IDE (ByteDance)	Open from terminal with `trae` command.
Continue	VS Code / JetBrains plugin	Config: `~/.continue/config.json`. Supports all major providers.

Quick Model Selection Reference

Tool	Flag/Config	Example
Claude Code	`--model`	`claude --model claude-sonnet-4-20250514`
Codex CLI	`--model`	`codex --model o4-mini`
Gemini CLI	`--model`	`gemini --model gemini-2.5-pro`
Aider	`--model`	`aider --model claude-3.5-sonnet`
Goose	config file	`goose configure` then select provider/model
Copilot CLI	N/A	Uses GitHub Copilot model
OpenCode	config file	Set in `opencode.json`

Additional Resources

references/tools-comparison.md — Detailed side-by-side feature comparison table across all CLI AI coding tools, covering capabilities, model support, pricing, and platform availability.
references/common-patterns.md — Common usage patterns shared across CLI AI tools: context management, multi-file editing, git integration, MCP server configuration, and model routing strategies.

Source: biodoia/biodoia-skills-marketplace — distributed by TomeVault.

安装命令

一键复制安装选择包管理器,点击右侧复制,粘贴到终端回车即可

npx skills add tomevault-io/skills-registry/biodoia--biodoia-skills-marketplace--cli-ai-tools -y

查看通用安装教程

查看 Skill 原文

兼容工具

Claude Code
Cursor
Cline
Codex
Windsurf
Gemini
OpenCode
Roo Code
AAide
+Augment
GitHub Copilot
OpenClaw
KKiro
Junie
WWarp
Goose
Trae
Kilo Code
OpenHands
Amp
Zencoder
CContinue
AIAider
CodeBuddy
CodeGeeX
Replit Agent

安全审计

83 / 100

社区

未声明 license/许可。
未声明 version。
未限定 allowed-tools，默认拥有全部工具权限。
检出高风险片段：pipe_curl_to_shell

元信息

来源: github
tomevault-io/skills-registry
English

@tomevault-io 的其他技能

分类与标签

分类 AI 智能

先判断是否适合

作者设计意图

作者的方法与取舍

边界和复核

CLI AI Coding Tools Reference

Tool Index

Anthropic: Claude Code

OpenAI: Codex CLI

Google: Gemini CLI

Google: Jules

Open Source: Aider

Block: Goose

Sourcegraph: amp

GitHub: Copilot CLI

Sourcegraph: Cody CLI

Open Source: OpenCode

Qodo: qodo

Neovim: avante.nvim

Other CLI-Adjacent Tools

Quick Model Selection Reference

Additional Resources

讨论