research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Research Skill

Quick Ref: Deep codebase exploration with multi-angle analysis. Output:
.agents/research/*.md

YOU MUST EXECUTE THIS WORKFLOW. Do not just describe it.

CLI dependencies: ao (knowledge injection — optional). If ao is unavailable, skip prior knowledge search and proceed with direct codebase exploration.

速查参考： 多维度分析的深度代码库探索。输出文件：
.agents/research/*.md

你必须执行此工作流，而非仅描述它。

CLI依赖： ao（知识注入——可选）。如果ao不可用，跳过前置知识搜索，直接进行代码库探索。

Execution Steps

执行步骤

Given

/research <topic>

当输入

/research <topic>

时：

Step 1: Create Output Directory

步骤1：创建输出目录

bash

mkdir -p .agents/research

bash

mkdir -p .agents/research

Step 2: Check Prior Art

步骤2：检查已有成果

First, search and inject existing knowledge (if ao available):

bash

undefined

首先，搜索并注入现有知识（若ao可用）：

bash

undefined

Search knowledge base for relevant learnings, patterns, and prior research

搜索知识库，查找相关研究成果、模式和过往研究

ao search "<topic>" 2>/dev/null || echo "ao not available, skipping knowledge search"

Inject relevant context into this session

将相关上下文注入当前会话

ao inject "<topic>" 2>/dev/null || echo "ao not available, skipping knowledge injection"


**Review ao search results:** If ao returns relevant learnings or patterns, incorporate them into your research strategy. Look for:
- Prior research on this topic or related topics
- Known patterns or anti-patterns
- Lessons learned from similar investigations

**Search local research artifacts:**
```bash
ls -la .agents/research/ 2>/dev/null | grep -i "<topic>" || echo "No prior research found"

Also use Grep to search

.agents/

for related content. Check TEMPERED learnings:

bash

ls -la .agents/learnings/ .agents/patterns/ 2>/dev/null | head -10

ao inject "<topic>" 2>/dev/null || echo "ao not available, skipping knowledge injection"


**查看ao搜索结果：** 如果ao返回相关研究成果或模式，将其纳入你的研究策略。重点关注：
- 该主题或相关主题的过往研究
- 已知的模式或反模式
- 类似调查中获得的经验教训

**搜索本地研究工件：**
```bash
ls -la .agents/research/ 2>/dev/null | grep -i "<topic>" || echo "No prior research found"

同时使用Grep搜索

.agents/

目录下的相关内容。检查TEMPERED研究成果：

bash

ls -la .agents/learnings/ .agents/patterns/ 2>/dev/null | head -10

Step 3: Launch Explore Agent

步骤3：启动Explore Agent

YOU MUST USE THE TASK TOOL NOW. Call it with these exact parameters:

Tool: Task
Parameters:
  subagent_type: "Explore"
  description: "Research: <topic>"
  prompt: |
    Thoroughly investigate: <topic>

    Search strategy:
    1. Glob for relevant files (*.md, *.py, *.ts, *.go, etc.)
    2. Grep for keywords related to <topic>
    3. Read key files and understand the architecture
    4. Check docs/ and .agents/ for existing documentation

    Return a detailed report with:
    - Key files found (with paths)
    - How the system works
    - Important patterns or conventions
    - Any issues or concerns

    Cite specific file:line references for all claims.

你现在必须使用Task工具。 调用时使用以下精确参数：

Tool: Task
Parameters:
  subagent_type: "Explore"
  description: "Research: <topic>"
  prompt: |
    彻底调查：<topic>

    搜索策略：
    1. 匹配相关文件（*.md, *.py, *.ts, *.go等）
    2. 使用Grep搜索与<topic>相关的关键词
    3. 阅读关键文件并理解架构
    4. 检查docs/和.agents/目录下的现有文档

    返回一份详细报告，包含：
    - 找到的关键文件（含路径）
    - 系统工作原理
    - 重要模式或约定
    - 任何问题或关注点

    所有结论都需标注`file:line`引用。

Step 4: Validate Research Quality (Optional)

步骤4：验证研究质量（可选）

For thorough research, perform quality validation:

若要进行全面研究，执行质量验证：

4a. Coverage Validation

4a. 覆盖范围验证

Check: Did we look everywhere we should? Any unexplored areas?

List directories/files explored
Identify gaps in coverage
Note areas that need deeper investigation

检查：我们是否查看了所有应检查的内容？是否有未探索的领域？

列出已探索的目录/文件
识别覆盖范围的缺口
标注需要深入调查的领域

4b. Depth Validation

4b. 深度验证

Check: Do we UNDERSTAND the critical parts? HOW and WHY, not just WHAT?

Rate depth (0-4) for each critical area
Flag areas with shallow understanding
Identify what needs more investigation

检查：我们是否理解了关键部分？不仅要知道是什么，还要知道如何运作以及为什么？

为每个关键领域的深度评分（0-4分）
标记理解较浅的领域
识别需要进一步调查的内容

4c. Gap Identification

4c. 缺口识别

Check: What DON'T we know that we SHOULD know?

List critical gaps
Prioritize what must be filled before proceeding
Note what can be deferred

检查：有哪些我们应该知道但不知道的内容？

列出关键知识缺口
优先排序必须填补的缺口，再继续后续工作
标注可延后处理的内容

4d. Assumption Challenge

4d. 假设验证

Check: What assumptions are we building on? Are they verified?

List assumptions made
Flag high-risk unverified assumptions
Note what needs verification

检查：我们基于哪些假设？这些假设是否已验证？

列出已做出的假设
标记高风险未验证假设
标注需要验证的内容

Step 5: Synthesize Findings

步骤5：整合研究结果

After the Explore agent and validation swarm return, write findings to:

.agents/research/YYYY-MM-DD-<topic-slug>.md

Use this format:

markdown

undefined

在Explore Agent和验证集群返回结果后，将研究结果写入：

.agents/research/YYYY-MM-DD-<topic-slug>.md

使用以下格式：

markdown

undefined

Research: <Topic>

Date: YYYY-MM-DD Scope: <what was investigated>

日期： YYYY-MM-DD 范围： <调查内容>

Summary

摘要

<2-3 sentence overview>

<2-3句话的概述>

Key Files

关键文件

File	Purpose
path/to/file.py	Description

文件	用途
path/to/file.py	描述

Findings

研究结果

<详细研究结果，含file:line引用>

Recommendations

建议

<next steps or actions> ```

<后续步骤或行动>

undefined

Step 6: Request Human Approval (Gate 1)

步骤6：请求人工批准（关卡1）

USE AskUserQuestion tool:

Tool: AskUserQuestion
Parameters:
  questions:
    - question: "Research complete. Approve to proceed to planning?"
      header: "Gate 1"
      options:
        - label: "Approve"
          description: "Research is sufficient, proceed to /plan"
        - label: "Revise"
          description: "Need deeper research on specific areas"
        - label: "Abandon"
          description: "Stop this line of investigation"
      multiSelect: false

Wait for approval before reporting completion.

使用AskUserQuestion工具：

Tool: AskUserQuestion
Parameters:
  questions:
    - question: "Research complete. Approve to proceed to planning?"
      header: "Gate 1"
      options:
        - label: "Approve"
          description: "Research is sufficient, proceed to /plan"
        - label: "Revise"
          description: "Need deeper research on specific areas"
        - label: "Abandon"
          description: "Stop this line of investigation"
      multiSelect: false

在获得批准前，不要报告完成。

Step 7: Report to User

步骤7：向用户报告

Tell the user:

What you found
Where the research doc is saved
Gate 1 approval status
Next step:
```
/plan
```
to create implementation plan

告知用户：

你发现了什么
研究文档的保存位置
关卡1的批准状态
下一步：使用
```
/plan
```
创建实施计划

Key Rules

核心规则

Actually dispatch the Explore agent - don't just describe doing it
Scope searches - use the topic to narrow file patterns
Cite evidence - every claim needs
```
file:line
```
Write output - research must produce a
```
.agents/research/
```
artifact

实际调用Explore Agent - 不要仅描述调用过程
限定搜索范围 - 使用主题缩小文件模式
引用证据 - 所有结论都需要
```
file:line
```
引用
生成输出文件 - 研究必须生成
```
.agents/research/
```
下的工件

Thoroughness Levels

详尽程度等级

Include in your Explore agent prompt:

"quick" - for simple questions
"medium" - for feature exploration
"very thorough" - for architecture/cross-cutting concerns

在你的Explore Agent提示中加入：

"quick" - 用于简单问题
"medium" - 用于功能探索
"very thorough" - 用于架构/跨领域问题