Back to Details

gemini-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Gemini Research Skill

Gemini 研究 Skill

LIBRARY-FIRST PROTOCOL (MANDATORY)

库优先协议（强制性）

Before writing ANY code, you MUST check:
在编写任何代码之前，你必须检查：

Step 1: Library Catalog

步骤1：库目录

  • Location: 
    .claude/library/catalog.json
  • If match >70%: REUSE or ADAPT
  • 位置：
    .claude/library/catalog.json
  • 如果匹配度>70%：复用或适配

Step 2: Patterns Guide

步骤2：模式指南

  • Location: 
    .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
  • If pattern exists: FOLLOW documented approach
  • 位置：
    .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
  • 如果模式存在：遵循文档化方法

Step 3: Existing Projects

步骤3：现有项目

  • Location: 
    D:\Projects\*
  • If found: EXTRACT and adapt
  • 位置：
    D:\Projects\*
  • 如果找到：提取并适配

Decision Matrix

决策矩阵

MatchAction
Library >90%REUSE directly
Library 70-90%ADAPT minimally
Pattern existsFOLLOW pattern
In projectEXTRACT
No matchBUILD (add to library after)
匹配度操作
库>90%直接复用
库70-90%最小化适配
模式存在遵循模式
在项目中存在提取
无匹配构建（之后添加到库）

Purpose

用途

Route research tasks to Gemini CLI when:
  • Real-time information is needed (Google Search grounding)
  • Context exceeds Claude's 200k limit (Gemini has 1M)
  • Need web-grounded factual answers
当出现以下情况时，将研究任务路由到Gemini CLI：
  • 需要实时信息（Google Search grounding）
  • 上下文超过Claude的20万token限制（Gemini支持100万token）
  • 需要基于网络的事实性答案

Unique Capability

独特能力

What Gemini Does Better:
  • Google Search grounding for current information
  • 1M token context for massive document analysis
  • 70+ extensions (Figma, Stripe, Shopify, etc.)
  • Web content analysis with source attribution
Gemini的优势:
  • 借助Google Search grounding获取当前信息
  • 100万token上下文用于大规模文档分析
  • 70+扩展插件（Figma、Stripe、Shopify等）
  • 带来源归因的网页内容分析

When to Use

使用场景

Perfect For:

适用场景:

  • Current events, recent documentation
  • Large codebase analysis (>150k tokens)
  • Literature reviews with many papers
  • Real-time API documentation lookup
  • Market research, competitor analysis
  • 当前事件、最新文档
  • 大型代码库分析（>15万token）
  • 多文献综述
  • 实时API文档查询
  • 市场研究、竞品分析

Don't Use When:

不适用场景:

  • Offline/airgapped environments
  • Complex multi-step reasoning (use Claude)
  • Code generation requiring iteration (use Codex)
  • 离线/气隙环境
  • 复杂多步骤推理（使用Claude）
  • 需要迭代的代码生成（使用Codex）

Usage

使用方法

Basic Research

基础研究

bash
/gemini-research "What are the latest React 19 best practices?"
bash
/gemini-research "What are the latest React 19 best practices?"

With Context Files

带上下文文件

bash
/gemini-research "Analyze architecture" --context @src/
bash
/gemini-research "Analyze architecture" --context @src/

Large Document Analysis

大型文档分析

bash
/gemini-research "Summarize all papers" --context papers/*.pdf
bash
/gemini-research "Summarize all papers" --context papers/*.pdf

Command Pattern

命令模式

bash
bash scripts/multi-model/gemini-research.sh "<query>" "<task_id>" "json"
bash
bash scripts/multi-model/gemini-research.sh "<query>" "<task_id>" "json"

Memory Integration

内存集成

Results stored to Memory-MCP:
  • Key: 
    multi-model/gemini/research/{task_id}
  • Tags: WHO=gemini-cli, WHY=research
结果将存储到Memory-MCP:
  • 键: 
    multi-model/gemini/research/{task_id}
  • 标签: WHO=gemini-cli, WHY=research

Output Format

输出格式

json
{
  "content": "Research findings...",
  "sources": ["url1", "url2"],
  "model": "gemini-2.5-pro",
  "timestamp": "2025-12-28T..."
}
json
{
  "content": "Research findings...",
  "sources": ["url1", "url2"],
  "model": "gemini-2.5-pro",
  "timestamp": "2025-12-28T..."
}

Handoff to Claude

交接给Claude

After Gemini research completes:
  1. Results stored in Memory-MCP
  2. Claude agents read from memory key
  3. Use research to inform implementation
javascript
// Claude agent reads Gemini research
const research = memory_retrieve("multi-model/gemini/research/{task_id}");
Task("Coder", `Implement using: ${research.content}`, "coder");
Gemini研究完成后:
  1. 结果存储到Memory-MCP
  2. Claude代理从内存键读取数据
  3. 使用研究结果指导实现
javascript
// Claude agent reads Gemini research
const research = memory_retrieve("multi-model/gemini/research/{task_id}");
Task("Coder", `Implement using: ${research.content}`, "coder");

Configuration

配置

  • Retries: 3 attempts on failure
  • Timeout: 60 seconds per query
  • Fallback: Claude researcher agent if Gemini unavailable
  • 重试次数: 失败时尝试3次
  • 超时时间: 每次查询60秒
  • 降级方案: 如果Gemini不可用，使用Claude研究代理