Token导航 LogoToken导航

Agent Skill · 研究检索

gemini-chat(Gemini 聊天)

gemini-chat 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

gemini-chat 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为调用外部服务;许可证为 MIT,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 55,572。

总安装 474周安装 19GitHub Stars 9下载 154核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:gemini-chat(Gemini 聊天)
来源仓库:https://github.com/adaptationio/skrillz
仓库路径:skills/gemini-chat
安装命令:
npx skills add https://github.com/adaptationio/skrillz --skill gemini-chat
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/adaptationio/skrillz --skill gemini-chat

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

gemini-chat 支持交互式对话,集成 Google Gemini 模型,具备上下文管理和会话持久化能力。适合用于复杂多步任务、代码重构或多模态内容处理场景。支持实时切换 YOLO 模式以启用自动批准和执行,提升自动化程度。可通过命令行启动会话,并可选择包含特定目录或启用检查点机制。

需注意 OAuth 认证流程及网络依赖,确保 API 密钥或账户权限有效。

来源仓库内路径:skills/gemini-chat

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/adaptationio/skrillz --skill gemini-chat;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

gemini-chat 怎么安装?

在终端执行 npx skills add https://github.com/adaptationio/skrillz --skill gemini-chat。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

gemini-chat 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

gemini-chat 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

gemini-chat 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

gemini-chat 用的人多吗?

公开数据是累计安装 474 次,近一周安装 19 次,下载 154 次,GitHub 9 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

gemini-chat 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。

先想清楚的地方

  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
gemini-chat

Gemini 聊天 · 本页

4749调用外部服务MIT有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Gemini Interactive Chat

Advanced interactive chat workflows with Gemini CLI, featuring context management, multimodal support, and session persistence.

Quick Start

# Start interactive chat
gemini -i "Let's work on my project"

# Start with specific context
gemini --include-directories ./src -i "Help me refactor this code"

# Start with checkpointing and YOLO mode (toggle in session)
gemini --checkpointing -i "Complex multi-step task"

YOLO Mode in Interactive Sessions

YOLO mode can be toggled during interactive sessions for automated execution:

# In interactive mode:
# Ctrl+Y              # Toggle YOLO mode on/off
# /yolo on            # Enable auto-approval
# /yolo off           # Disable auto-approval

# Start with YOLO already enabled (for automation)
gemini --yolo -i "Set up complete development environment automatically"

# Mixed mode workflow
gemini -i "Let's plan first, then execute with YOLO"
# Plan manually, then: Ctrl+Y to enable YOLO for execution

When to use YOLO in chat:

  • ✅ File operations (create, edit, organize)
  • ✅ Batch processing tasks
  • ✅ Project setup and configuration
  • ✅ Documentation generation
  • ❌ Complex system changes (review first)
  • ❌ Unknown/experimental commands

Interactive Commands

Essential Commands

/help              # Show all available commands
/tools             # List available tools
/mcp               # Show MCP server status
/compress          # Compress conversation context
/copy              # Copy last response
/clear             # Clear screen and context
/checkpoint        # Save current state
/restore           # Restore previous state
/exit              # End session

Context Management

# In chat, reference files
@./src/main.js     # Include file in context
@./docs/           # Include entire directory
@https://example.com  # Include web page

# Include images
@./diagram.png     # Add image to conversation
Ctrl+V             # Paste image from clipboard

# Context commands
/compress          # Summarize long conversation
/clear context     # Clear context, keep chat
/show context      # Display current context

Chat Workflows

Code Review Session

#!/bin/bash
# Interactive code review

start_code_review() {
  local project_dir="${1:-.}"

  gemini \
    --include-directories "$project_dir/src" \
    --exclude-directories node_modules,.git \
    --checkpointing \
    -i "Let's do a code review. Start by giving me an overview of the codebase structure."
}

# Usage
start_code_review ./my-project

Brainstorming Session

#!/bin/bash
# Creative brainstorming with saves

brainstorm_session() {
  local topic="$1"
  local session_name="brainstorm-$(date +%Y%m%d-%H%M%S)"

  # Create session directory
  mkdir -p ~/.gemini/sessions/$session_name

  # Start session with logging
  gemini \
    --checkpointing \
    --log ~/.gemini/sessions/$session_name/chat.log \
    -i "Let's brainstorm about: $topic

Please help me:
    1. Generate creative ideas
    2. Evaluate feasibility
    3. Identify challenges
    4. Suggest next steps"

  # Save final summary with YOLO
  gemini --yolo -p "Summarize our brainstorming session and create actionable next steps" \
    > ~/.gemini/sessions/$session_name/summary.md

  echo "Session saved to ~/.gemini/sessions/$session_name/"
}

# Automated brainstorming with YOLO
automated_brainstorm() {
  local topic="$1"
  local session_name="auto-brainstorm-$(date +%Y%m%d-%H%M%S)"

  mkdir -p ~/.gemini/sessions/$session_name

  # Fully automated brainstorming session
  gemini --yolo --checkpointing \
    -i "Automated brainstorming session for: $topic

Execute this workflow automatically:
1. Generate 10 creative ideas
2. Evaluate each for feasibility (1-10 scale)
3. Identify top 3 ideas with rationale
4. Create implementation roadmap for top idea
5. List potential challenges and solutions
6. Generate final summary document

Save all outputs to organized files."
}

Learning Session

#!/bin/bash
# Interactive learning with examples

learn_with_gemini() {
  local subject="$1"

  gemini \
    --yolo \
    -i "I want to learn about $subject.

    Please:
    1. Start with fundamentals
    2. Provide examples
    3. Create exercises
    4. Check my understanding
    5. Suggest resources

    Let's begin!"
}

Advanced Features

Session Management

#!/bin/bash
# Save and resume sessions

# Save session
save_session() {
  local name="${1:-session}"
  gemini checkpoint save "$name"
  echo "Session saved as: $name"
}

# List sessions
list_sessions() {
  gemini checkpoint list
}

# Resume session
resume_session() {
  local name="$1"
  gemini checkpoint restore "$name"
  gemini -i "Let's continue where we left off"
}

# Delete old sessions
cleanup_sessions() {
  local days_old="${1:-7}"
  find ~/.gemini/checkpoints -mtime +$days_old -delete
  echo "Cleaned up sessions older than $days_old days"
}

Multi-Modal Conversations

#!/bin/bash
# Work with images and text

analyze_designs() {
  local design_dir="$1"

  gemini -i "Let's review UI designs"

  # In the chat:
  # @./designs/homepage.png
  # @./designs/dashboard.png
  # "Compare these designs and suggest improvements"
}

# Screenshot analysis
analyze_screenshot() {
  # Take screenshot (macOS)
  screencapture -i /tmp/screenshot.png

  # Or Linux
  # import /tmp/screenshot.png

  gemini -p "Analyze this screenshot: @/tmp/screenshot.png"
}

Context Optimization

#!/bin/bash
# Manage context efficiently

optimized_chat() {
  local max_context="${1:-50000}"  # tokens

  # Monitor context size
  while true; do
    context_size=$(gemini context size)

    if [ $context_size -gt $max_context ]; then
      echo "Context limit approaching. Compressing..."
      gemini /compress
    fi

    # Continue chat
    gemini -i "Continue"
  done
}

# Selective context
focused_chat() {
  local focus_dir="$1"
  local task="$2"

  # Start with minimal context
  gemini \
    --include-directories "$focus_dir" \
    --max-depth 1 \
    -i "Focus only on: $task"
}

Prompt Engineering

Structured Prompts

# Engineering prompt
gemini -i "You are a senior software engineer.

Context:
- Language: Python
- Framework: FastAPI
- Database: PostgreSQL

Task: Help me design a scalable API

Constraints:
- Must handle 10k requests/second
- Must have <100ms response time
- Must be maintainable

Let's start with the architecture."

Role-Based Sessions

#!/bin/bash
# Different personas for different tasks

start_role_session() {
  local role="$1"

  case $role in
    architect)
      PROMPT="You are a solutions architect. Focus on system design, scalability, and best practices."
      ;;
    reviewer)
      PROMPT="You are a code reviewer. Focus on bugs, security issues, and code quality."
      ;;
    teacher)
      PROMPT="You are a patient teacher. Explain concepts simply with examples."
      ;;
    debugger)
      PROMPT="You are a debugging expert. Focus on finding and fixing issues systematically."
      ;;
    *)
      PROMPT="You are a helpful AI assistant."
      ;;
  esac

  gemini -i "$PROMPT

  How can I help you today?"
}

Automation Scripts

Daily Standup

#!/bin/bash
# Automated daily standup

daily_standup() {
  local git_log=$(git log --oneline -5)
  local todos=$(cat TODO.md 2>/dev/null || echo "No TODO file")

  gemini -p "Based on recent commits and TODOs, generate a standup update:

  Recent commits:
  $git_log

  TODOs:
  $todos

  Format:
  - What I did yesterday
  - What I'm doing today
  - Any blockers"
}

Continuous Learning

#!/bin/bash
# Learn from codebase changes

learn_from_changes() {
  local since="${1:-1 week ago}"

  # Get changes
  local changes=$(git diff --stat "@{$since}")
  local new_files=$(git ls-files --others --exclude-standard)

  gemini --yolo -i "Analyze these recent changes and teach me:

  Changes since $since:
  $changes

  New files:
  $new_files

  Please:
  1. Explain what was changed and why
  2. Identify patterns and best practices
  3. Suggest improvements
  4. Generate documentation for changes
  5. Create knowledge summary document
  6. Quiz me on the concepts"
}

# Automated development workflow session
auto_dev_workflow() {
  local task="$1"

  gemini --yolo --checkpointing -i "Automated development workflow: $task

Execute this complete development cycle:
1. Analyze current codebase and requirements
2. Create detailed implementation plan
3. Generate all necessary code files
4. Write comprehensive tests
5. Create documentation
6. Set up CI/CD integration
7. Generate deployment scripts
8. Create progress report

Work autonomously and save all outputs."
}

Integration Patterns

Vim Integration

" .vimrc
" Send selection to Gemini
vnoremap <leader>g :w !gemini -p "Explain this code: $(cat)"<CR>

" Get suggestions
nnoremap <leader>G :r !gemini -p "Improve this function: $(cat %:p)"<CR>

" Interactive help
nnoremap <leader>? :terminal gemini -i "Help with vim and %:t"<CR>

VS Code Integration

# Connect Gemini to VS Code
gemini /ide install
gemini /ide enable

# Now Gemini can see your VS Code context
gemini -i "Help me with the file I have open in VS Code"

Tmux Workflow

#!/bin/bash
# Split-screen coding with Gemini

gemini_tmux() {
  tmux new-session -d -s gemini-coding

  # Main coding pane
  tmux send-keys -t gemini-coding "vim" C-m

  # Gemini chat pane
  tmux split-window -h -t gemini-coding
  tmux send-keys -t gemini-coding "gemini -i 'Help me code'" C-m

  # File tree pane
  tmux split-window -v -t gemini-coding
  tmux send-keys -t gemini-coding "watch -n 2 'tree -L 2'" C-m

  # Attach to session
  tmux attach -t gemini-coding
}

Best Practices

Effective Prompting

  1. Be Specific # Bad "Help me with code" # Good "Help me refactor this authentication module to use JWT tokens"
  2. Provide Context # Include relevant files gemini --include-directories./src/auth,./tests/auth
  3. Set Constraints "Explain in simple terms, with examples, max 100 lines"

Session Hygiene

  1. Regular Compression # Compress every 30 minutes /compress
  2. Save Important Sessions /checkpoint
  3. Clear Irrelevant Context /clear context

Troubleshooting

Common Issues

  1. Slow Responses

- Compress context: /compress - Use faster model: gemini -m gemini-2.5-flash - Reduce included directories

  1. Lost Context

- Check checkpoints: /restore - Review history: ~/.gemini/history

  1. Tool Errors

- Check available tools: /tools - Verify permissions: ls -la

Related Skills

  • gemini-cli: Main Gemini CLI integration
  • gemini-auth: Authentication management
  • gemini-tools: Tool execution workflows
  • gemini-mcp: MCP server management

平台分布

github-copilot

30.63%
按下载量换算47

Claude Code

21.86%
按下载量换算34

mcpjam

18.3%
按下载量换算28

moltbot

12.84%
按下载量换算20

windsurf

7.71%
按下载量换算12

zencoder

3.25%
按下载量换算5

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。

继续浏览同类 Skills