Getting Started with Claude GitHub

Install the full repository optimization suite and run your first audit in under 5 minutes. This guide covers prerequisites, installation, optional integrations, and the recommended workflow.

Prerequisites

Claude GitHub requires three tools installed on your system:

• Claude Code CLI - the command-line interface for Claude. Install from claude.ai/code or via npm install -g @anthropic-ai/claude-code.
• GitHub CLI (gh) - used by audit agents to fetch repository metadata. Install from cli.github.com.
• Node.js 18+ - required for the DataForSEO MCP server. Download from nodejs.org.

The installer validates all prerequisites and will tell you if anything is missing.

Installation

git clone https://github.com/avalonreset/claude-github.git
cd claude-github

bash install.sh

.\install.ps1

The installer copies all 8 skill files to ~/.claude/skills/github/, sets up the DataForSEO MCP server configuration, and optionally configures KIE.ai for AI-generated banner images. After installation, restart Claude Code and the skills are available immediately.

Optional: DataForSEO Setup

DataForSEO provides live keyword data including monthly search volume, keyword difficulty scores, and SERP competitor analysis. It is optional - all skills work without it by falling back to codebase analysis and built-in reference guides.

If you want live keyword data:

1. Create an account at dataforseo.com
2. The installer prompts for your API login and password during setup
3. Credentials are stored in ~/.claude/.mcp.json
4. A typical single-repo audit costs 15-30 cents in API credits

Without DataForSEO, keyword recommendations from /github seo will be marked as unverified. All other skills function identically. Read more about keyword research in the GitHub SEO keywords guide.

Optional: KIE.ai Setup

KIE.ai generates AI banner images for README files through the /github readme command. This is purely optional and cosmetic. The installer prompts for KIE.ai credentials if you want to enable it.

Your First Audit

Open any GitHub repository in Claude Code and run:

/github audit

This spawns 6 parallel scoring agents that evaluate your repository across README Quality (25%), Metadata and Discovery (20%), Legal Compliance (15%), Community Health (15%), Release and Maintenance (15%), and SEO and Discoverability (10%). The output includes a 0-100 health score, a gap analysis, and a numbered remediation plan.

For a detailed explanation of what each category measures and how to interpret the results, read the audit walkthrough guide.

The Standard Operating Procedure

After your first audit, follow the remediation plan in this order:

1. /github legal - select a license, generate SECURITY.md and CITATION.cff
2. /github community - create issue templates, CONTRIBUTING.md, CODE_OF_CONDUCT.md
3. /github release - set up CHANGELOG, badges, and semantic versioning
4. /github seo - research keywords with real search volume data
5. /github meta - optimize description, topics, and feature toggles
6. /github readme - rewrite README using keyword data and all generated files
7. /github audit - re-audit to measure improvement

Each command generates production-ready files that are ready to commit immediately. The workflow section on the homepage has a visual overview of this process.

Command Reference

Portfolio Audits

To audit all public repositories under a GitHub account, run:

/github audit username

This fetches all public repos, ranks them by activity and stars, and runs deep audits on the top candidates. The output is a cross-portfolio report with shared patterns and a prioritized remediation plan across all repositories.

After fixing individual repos, use /github empire to optimize your GitHub profile as a whole, including a profile README, cross-linking strategy, topic synchronization, and AI-generated avatar.