/gitingest

Purpose: Git repository analysis and markdown digest generation for LLM ingestion using GitHub MCP tools

---

@include shared/universal-constants.yml#Universal_Legend

Command Execution

Execute: immediate. --plan→show plan first Legend: Generated based on symbols used in command Purpose: "[Action][Subject] - Analyze GitHub repository and create markdown digest"

Analyze GitHub repositories using native MCP Docker GitHub tools and create LLM-optimized markdown digests automatically saved to your Obsidian vault.

@include shared/flag-inheritance.yml#Universal_Always

Usage

/gitingest <github-url> [options]

Examples:

• /gitingest https://github.com/user/repo - Analyze public GitHub repo
• /gitingest https://github.com/user/private-repo - Analyze private repo (uses your authenticated credentials)
• /gitingest https://github.com/user/repo --branch feature-branch - Analyze specific branch
• /gitingest https://github.com/user/repo --focus code - Focus on code files only

Parameters

Required:

• <github-url>: Full GitHub repository URL (e.g., https://github.com/owner/repo)

Optional:

• --branch <name>: Specific branch to analyze (default: repo's default branch)
• --focus <mode>: Analysis focus mode
  • code: Code files only (*.js, *.py, *.ts, *.go, *.java, etc.)
  • docs: Documentation only (*.md, *.txt, *.rst, etc.)
  • config: Configuration files (*.json, *.yml, *.toml, Dockerfile, etc.)
  • all: All files (default)

Implementation Details

This command uses MCP Docker GitHub tools:

1. Repository Access: mcp__MCP_DOCKER__get_file_contents to browse repository
2. File Reading: Recursive file tree traversal
3. Content Extraction: Direct file content retrieval from GitHub API
4. Metadata Collection: Repository info, commits, branches

Authentication:

• Uses your existing GitHub credentials configured in MCP Docker
• Works with both public and private repositories
• No additional token setup required

Output:

• Auto-saves to: /Users/zorro/Documents/Obsidian/Claudecode/
• Filename format: YYYY-MM-DD-<repo-name>-repository-analysis.md
• Includes frontmatter with tags, metadata, and classification

Analysis Process

The command will:

1. Parse GitHub URL to extract owner and repository name
2. Fetch repository structure using get_file_contents with path="/"
3. Filter files based on --focus mode (if specified)
4. Read file contents for all relevant files
5. Collect metadata:
   • Recent commits (last 10)
   • Branch information
   • Repository description
6. Generate markdown digest with:
   • Repository overview
   • File tree structure
   • Key file contents
   • Technical stack analysis
   • Usage examples (from README if available)
   • Metadata and tags

Focus Modes

Code Focus (--focus code):

• Includes: *.js, *.ts, *.py, *.go, *.java, *.rb, *.php, *.rs, *.c, *.cpp, *.h, *.swift, *.kt
• Excludes: Documentation, configs, build artifacts

Docs Focus (--focus docs):

• Includes: *.md, *.txt, *.rst, *.adoc, CHANGELOG, LICENSE
• Excludes: Code files, configs

Config Focus (--focus config):

• Includes: *.json, *.yml, *.yaml, *.toml, *.ini, Dockerfile, .env.example, package.json, requirements.txt
• Excludes: Code, documentation

All Files (--focus all, default):

• Includes everything except common excludes:
  • node_modules/, .git/, dist/, build/, __pycache__/
  • Binary files (detected by file extension)
  • Files larger than 1MB

Output Format

The generated markdown file includes:

---
title: "<Repo Name>: <Brief Description>"
tags:
  - repository
  - <detected-languages>
  - <tech-stack>
  - technical
  - inbox
url: <github-url>
date: <capture-date>
type: repository
status: inbox
priority: medium
owner: <github-owner>
visibility: <public|private>
---

# [Repository Name](github-url)

## 📖 Repository Overview
<Auto-generated description>

## 📁 Repository Structure
<File tree>

## 🔑 Key Files

### <filename>
<File contents or summary>

## 🛠️ Technical Stack
<Detected technologies>

## 📊 Repository Metrics
<Commits, branches, size, etc.>

## 🏷️ Tags Analysis
<Why these tags were chosen>

Use Cases

• Code Review Preparation: Get full context before reviewing PRs
• Architecture Analysis: Understand project structure quickly
• Onboarding: Help new team members understand codebase
• Documentation: Generate knowledge base entries
• AI Context: Prepare repository context for Claude Code analysis

Limitations

• File Size: Files over 1MB are summarized (not included in full)
• Binary Files: Images, PDFs, and other binaries are listed but not included
• Large Repositories: Repos with 500+ files may be sampled
• Rate Limits: GitHub API rate limits apply (60 req/hour unauthenticated, 5000/hour authenticated)

Error Handling

• 404 Not Found: Repository doesn't exist or is private (and you lack access)
• 403 Forbidden: Rate limit exceeded or insufficient permissions
• Invalid URL: Provide full GitHub URL format

@include shared/research-patterns.yml#Mandatory_Research_Flows

@include shared/docs-patterns.yml#Standard_Notifications

@include shared/universal-constants.yml#Standard_Messages_Templates