github-repo-explore

bởi microsoft

Hướng dẫn tìm kiếm và khám phá các kho lưu trữ GitHub bên ngoài bằng cách sử dụng CLI gh. Sử dụng khi bạn cần các triển khai tham khảo, mẫu thiết kế hoặc ví dụ mã nguồn…

npx skills add https://github.com/microsoft/semantic-link-labs --skill github-repo-explore

GitHub Repository Search and Exploration

This skill enables agents to discover, clone, and explore external GitHub repositories when implementing features that may benefit from reference implementations in open-source projects.

When to Use This Skill

Use this skill when you need to:

  • Find reference implementations for features you're building
  • Explore how other projects implement similar patterns
  • Search for code examples across GitHub
  • Clone external repos for local exploration and learning
  • Find libraries or tools that could inform your implementation

Do NOT use this skill when:

  • The task is straightforward and doesn't need external references
  • The codebase already has established patterns to follow
  • You're working on proprietary code that shouldn't reference external sources

Resource Directory

All external GitHub repositories MUST be stored in:

.agent_cache/resources/

This is the centralized location for all cloned external repositories. Always check here first before searching GitHub.

Directory Structure

.agent_cache/
└── resources/
    ├── microsoft/
    │   ├── semantic-kernel/
    │   └── TypeChat/
    ├── langchain-ai/
    │   └── langchain/
    └── Significant-Gravitas/
        └── AutoGPT/

Repositories are organized by {owner}/{repo} structure to prevent naming conflicts.


Workflow Overview

┌─────────────────────────────────────────────────────────────────┐
│  1. Check Local Cache (.agent_cache/resources/)                 │
│     └─ If found → Use local repo for exploration                │
└────────────────────────────┬────────────────────────────────────┘
                             │ Not found
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│  2. Authenticate with GitHub (if needed)                        │
│     └─ gh auth login                                            │
└────────────────────────────┬────────────────────────────────────┘
                             │
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│  3. Search GitHub for Repositories                              │
│     └─ gh search repos <query>                                  │
│     └─ gh search code <query>                                   │
└────────────────────────────┬────────────────────────────────────┘
                             │ Found match
                             ▼
┌───────────────────────────────────────────────────────────────────┐
│  4. Clone to Cache Directory                                      │
│     └─ gh repo clone owner/repo .agent_cache/resources/owner/repo │
└────────────────────────────┬──────────────────────────────────────┘
                             │
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│  5. Explore Repository                                          │
│     └─ Use gh commands and file reading tools                   │
└─────────────────────────────────────────────────────────────────┘

Phase 1: Check Local Cache First

Before searching GitHub, always check if a relevant repository is already cloned locally.

List Cached Repositories

# List all cached repositories
ls -la .agent_cache/resources/

# List repos by owner
ls -la .agent_cache/resources/microsoft/

# Search for specific repo by name
find .agent_cache/resources -type d -name "*semantic*" 2>/dev/null

Explore Local Repository

If the repository exists locally, use standard file tools:

# View repository structure
tree -L 2 .agent_cache/resources/microsoft/semantic-kernel/

# Search for patterns in local repo
grep -r "planning" .agent_cache/resources/microsoft/semantic-kernel/python/ --include="*.py" | head -20

# Find specific files
find .agent_cache/resources/microsoft/semantic-kernel -name "*.md" -path "*skills*"

Phase 2: GitHub Authentication

If you need to search or clone from GitHub, ensure authentication is set up.

Check Authentication Status

gh auth status

Interactive Login (if needed)

# Interactive login - user will authenticate via browser
gh auth login

This opens a browser for the user to authenticate with their GitHub account.

Verify Login

# Verify authentication worked
gh auth status

Phase 3: Search GitHub

Use gh search commands to find relevant repositories or code.

Search for Repositories

# Search repos by keywords
gh search repos "agent planning skills" --limit 10

# Search with language filter
gh search repos "agent planning" --language python --limit 10

# Search by owner
gh search repos --owner microsoft "semantic kernel" --limit 10

# Search with JSON output for parsing
gh search repos "agent planning" --json fullName,description,stargazersCount --limit 5

# Sort by stars (most popular)
gh search repos "agent planning" --sort stars --limit 10

# Filter by number of stars
gh search repos "agent framework" --stars ">1000" --limit 10

Search for Code

# Search code across GitHub
gh search code "TaskLedger" --limit 10

# Search code in specific language
gh search code "def plan" --language python --limit 10

# Search code in specific repo
gh search code "planning" --repo microsoft/semantic-kernel --limit 20

# Search code by filename
gh search code "skills" --filename "*.md" --limit 10

# Search with JSON output
gh search code "agent skills" --json path,repository --limit 10

View Repository Details

# View repo README and description
gh repo view microsoft/semantic-kernel

# View repo with specific fields as JSON
gh repo view microsoft/semantic-kernel --json name,description,stargazerCount,defaultBranchRef

# Open repo in browser
gh repo view microsoft/semantic-kernel --web

Phase 4: Clone Repository to Cache

When you find a useful repository, clone it to the cache directory.

Clone Commands

# Create cache directory structure
mkdir -p .agent_cache/resources/{owner}

# Clone repository
gh repo clone microsoft/semantic-kernel .agent_cache/resources/microsoft/semantic-kernel

# Clone with shallow history (faster, uses less disk)
gh repo clone microsoft/semantic-kernel .agent_cache/resources/microsoft/semantic-kernel -- --depth 1

# Clone specific branch
gh repo clone microsoft/semantic-kernel .agent_cache/resources/microsoft/semantic-kernel -- --branch main --depth 1

Update Existing Clone

# Pull latest changes
cd .agent_cache/resources/microsoft/semantic-kernel && git pull && cd -

Phase 5: Explore Repository

Once cloned, use a combination of gh commands and file reading tools.

Repository Structure Exploration

# View directory structure
tree -L 3 .agent_cache/resources/microsoft/semantic-kernel/python/ | head -50

# Find README files
find .agent_cache/resources/microsoft/semantic-kernel -name "README.md" | head -10

# Find Python files related to agents
find .agent_cache/resources/microsoft/semantic-kernel -path "*agents*" -name "*.py" | head -20

Code Search in Local Clone

# Search for specific patterns
grep -r "class.*Agent" .agent_cache/resources/microsoft/semantic-kernel/python --include="*.py" | head -20

# Find function definitions
grep -rn "def plan" .agent_cache/resources/microsoft/semantic-kernel/python --include="*.py"

# Search for imports
grep -r "from.*import" .agent_cache/resources/microsoft/semantic-kernel/python --include="*.py" | grep planning

Read Specific Files

After identifying relevant files with grep/find, use the read_file tool to examine content:

# Example files to read
read_file .agent_cache/resources/microsoft/semantic-kernel/python/semantic_kernel/agents/orchestration/magentic.py

Useful gh Commands Reference

Repository Commands

CommandDescription
gh repo view <owner/repo>View repo README and details
gh repo view <owner/repo> --json <fields>Get repo metadata as JSON
gh repo clone <owner/repo> <path>Clone repo to specific path
gh repo list <owner>List repos owned by user/org

Search Commands

CommandDescription
gh search repos <query>Search for repositories
gh search code <query>Search within code
gh search issues <query>Search issues
gh search prs <query>Search pull requests

Search Filters

FilterExampleDescription
--language--language pythonFilter by programming language
--owner--owner microsoftFilter by repository owner
--stars--stars ">1000"Filter by star count
--limit--limit 20Limit number of results
--sort--sort starsSort results (stars, forks, updated)
--json--json fullName,descriptionOutput as JSON

JSON Fields for Repos

Common fields for --json output:

  • fullName - owner/repo
  • name - repo name only
  • description - repo description
  • stargazersCount - number of stars
  • language - primary language
  • url - GitHub URL
  • defaultBranchRef - main branch name

Example Workflows

Example 1: Find Reference Implementations for REST API Patterns

# 1. Check local cache
ls .agent_cache/resources/ 2>/dev/null || echo "Cache empty"

# 2. Search GitHub for Fabric/Power BI SDK examples
gh search repos "fabric rest api python" --stars ">50" --limit 10

# 3. View promising repo
gh repo view microsoft/semantic-link-labs

# 4. Clone to cache
mkdir -p .agent_cache/resources/microsoft
gh repo clone microsoft/semantic-link-labs .agent_cache/resources/microsoft/semantic-link-labs -- --depth 1

# 5. Explore API patterns
grep -r "_base_api" .agent_cache/resources/microsoft/semantic-link-labs --include="*.py" | head -20

Example 2: Search for Code Patterns

# Search for specific pattern across GitHub
gh search code "FabricRestClient" --language python --limit 5

# Clone repo containing the pattern
gh repo clone owner/repo .agent_cache/resources/owner/repo -- --depth 1

# Find and read the file
grep -rn "FabricRestClient" .agent_cache/resources/owner/repo --include="*.py"

Example 3: Explore Multiple Repos

# Search for agent frameworks
gh search repos "agent framework" --language python --stars ">2000" --json fullName,description,stargazersCount

# Clone top candidates
for repo in "langchain-ai/langchain" "Significant-Gravitas/AutoGPT"; do
    owner=$(dirname $repo)
    mkdir -p ".agent_cache/resources/$owner"
    gh repo clone "$repo" ".agent_cache/resources/$repo" -- --depth 1
done

# Compare implementations
diff <(grep -r "def plan" .agent_cache/resources/langchain-ai/langchain --include="*.py" | head -5) \
     <(grep -r "def plan" .agent_cache/resources/Significant-Gravitas/AutoGPT --include="*.py" | head -5)

Best Practices

  1. Always check local cache first -- Avoid redundant cloning
  2. Use shallow clones -- Add -- --depth 1 to save disk space and time
  3. Organize by owner/repo -- Prevents naming conflicts
  4. Use JSON output -- Easier to parse programmatically
  5. Start with grep/find -- Narrow down before reading full files
  6. Respect rate limits -- GitHub has API rate limits; cache results
  7. Clean up periodically -- Remove unused repos from cache

Troubleshooting

Authentication Issues

# Check current auth status
gh auth status

# Re-authenticate
gh auth login

# Use different auth method
gh auth login --with-token < token.txt

Rate Limiting

# Check rate limit status
gh api rate_limit

# If rate limited, wait or use cached repos

Clone Failures

# If clone fails, try with HTTPS explicitly
git clone https://github.com/owner/repo.git .agent_cache/resources/owner/repo

# Or try shallow clone
git clone --depth 1 https://github.com/owner/repo.git .agent_cache/resources/owner/repo

Thêm skills từ microsoft

oss-growth
microsoft
Cá tính tăng trưởng OSS
official
winapp-ui-automation
microsoft
Kiểm tra và tương tác với giao diện ứng dụng Windows đang chạy từ dòng lệnh bằng UI Automation (UIA). Sử dụng khi một tác nhân AI hoặc nhà phát triển cần kiểm tra giao diện người dùng...
official
accessibility-aria-expert
microsoft
Phát hiện và sửa các vấn đề về khả năng tiếp cận trong giao diện web React/Fluent UI. Sử dụng khi xem xét mã để đảm bảo tương thích với trình đọc màn hình, sửa nhãn ARIA, đảm bảo…
official
generate-canvas-app
microsoft
[DEPRECATED — sử dụng canvas-app thay thế] Tạo một ứng dụng canvas Power Apps hoàn chỉnh.
official
django
microsoft
Các phương pháp tốt nhất cho phát triển web Django bao gồm models, views, templates và testing.
official
github-issue-creator
microsoft
Chuyển đổi ghi chú thô, nhật ký lỗi, ghi âm giọng nói hoặc ảnh chụp màn hình thành báo cáo vấn đề markdown sắc nét theo phong cách GitHub. Sử dụng khi người dùng dán thông tin lỗi, lỗi…
official
python-package-management
microsoft
Sử dụng uv để quản lý phụ thuộc và poethepoet để tự động hóa tác vụ.
official
runtime-validation
microsoft
Xác thực thời gian chạy cho các ứng dụng đã di chuyển — bao gồm chiến lược kiểm thử (giai đoạn lập kế hoạch) và thực thi kiểm thử (giai đoạn xác thực): xác minh khởi động,…
official