Commands Audit Workflow Results FAQ Blog Docs Changelog GitHub
AUDITOR
README
SEO
LEGAL
Built for Claude Code // Repository Optimization

GitHub Repository Optimization Suite for Claude Code

Most GitHub repos have no keywords in the description, no structured README, missing license files, zero community health signals. Search engines skip them. Developers scroll past. Claude GitHub scores your repo 0-100 and generates every fix.

$ git clone https://github.com/avalonreset/claude-github.git && cd claude-github && bash install.sh

Works on macOS, Linux, Windows. Requires Claude Code + GitHub CLI.

All Commands How Scoring Works Avalon Reset
8
Skills
6
Parallel Agents
6
Scoring Categories
9
Reference Guides
<2min
Full Review
// What Is Claude GitHub

About the Claude GitHub Suite

What It Does

Claude GitHub is an open-source skill suite for the Claude Code CLI. It provides 8 specialized skills that analyze, score, and improve GitHub repositories across 6 weighted categories: README quality, metadata and discovery, legal compliance, community health, release and maintenance, and search discoverability. Not affiliated with Anthropic or GitHub, Inc.

How Scoring Works

When you run the /github audit command, the suite spawns 6 parallel scoring agents. Each one evaluates your project against detailed rubrics with specific point values, producing a 0-100 health score and a prioritized remediation plan. This analysis covers everything from README structure to search visibility. Every skill generates production-ready files including LICENSE, SECURITY.md, CONTRIBUTING.md, CITATION.cff, issue templates, CHANGELOG entries, and search-optimized README content.

Key Features

// How It Works

From Invisible to Discoverable

The Problem

Most GitHub projects are invisible to search engines and AI assistants. They have no keywords in the description, no structured README, missing license files, and zero community health files. Search engines cannot index what they cannot parse. AI systems like ChatGPT, Perplexity, and Google AI Overviews skip repositories that lack structured data.

The Fix

Claude GitHub fixes this in one pass. Run a 6-category assessment, identify every gap, and generate production-ready files for each issue. The workflow follows a standard operating procedure: diagnose with /github audit, research keywords with /github seo, generate fixes with targeted skills, then re-run the analysis to measure improvement.

Diagnose
Run /github audit
6 agents score your repository simultaneously across README quality, metadata, legal compliance, community health, release management, and search discoverability. You get a 0-100 score with a prioritized remediation plan in under 2 minutes.
Research
Live Keyword Data
The /github seo skill queries DataForSEO for real monthly search volumes, keyword difficulty scores, and competitor SERP analysis. Every recommendation is backed by actual search data, not assumptions about what developers might search for.
Fix
Generate Everything
Each skill generates production-ready files: LICENSE, SECURITY.md, CONTRIBUTING.md, issue templates, CHANGELOG, CI badges, and a search-optimized README. Nothing is left as a TODO. Every output is ready to commit.
Verify
Re-assess and Measure
Run the analysis again after applying fixes. Compare before and after scores across all 6 categories. Track improvement over time as you optimize your entire portfolio with /github empire.
// All Commands

8 Skills. One Suite.

Every recommendation cites its source: DataForSEO keyword volume, GitHub API metadata, codebase analysis, or reference guides. Nothing is guesswork. See the getting started guide for installation.

/github audit
Score any repository 0-100 across 6 categories with prioritized fixes. Spawns 6 agents in parallel.
/github legal
Select a license, generate SECURITY.md, CITATION.cff, handle fork compliance and attribution.
/github community
Generate issue templates, CONTRIBUTING.md, CODE_OF_CONDUCT.md, .gitattributes, CI workflow, devcontainer.
/github release
Plan release strategy, CHANGELOG, badges, semantic versioning, and package distribution.
/github seo
Run keyword research with real search volume and difficulty data via DataForSEO.
/github meta
Optimize description, topics, homepage URL, feature toggles, and social preview image.
/github readme
Generate or rewrite your README with search-optimized headings and AI-generated banner images.
/github empire
Portfolio strategy, profile README, AI avatar generation, cross-linking, and topic sync.
// How Scoring Works

0-100 Health Score

Parallel Agent Architecture

The scoring system uses 6 specialized agents that run in parallel. Each agent evaluates one category using a detailed rubric with specific point values, not subjective impressions. The final score is a weighted sum of all six categories.

Category Weights Explained

README Quality carries the highest weight at 25% because it is the primary surface developers, search engines, and AI assistants use to understand a project. Metadata and Discovery at 20% evaluates signals GitHub uses for search ranking and Explore recommendations. Legal Compliance, Community Health, and Release and Maintenance each carry 15%, covering files that signal project maturity. Search Discoverability at 10% checks keyword placement and AI citation readiness. After scoring, the tool generates a numbered remediation plan with the highest-impact fixes listed first.

  1. README Quality (25%) - structure, headings, badges, table of contents, code examples
  2. Metadata and Discovery (20%) - description keywords, topics, homepage URL, feature toggles
  3. Legal Compliance (15%) - license file, SECURITY.md, CITATION.cff, fork obligations
  4. Community Health (15%) - issue templates, CONTRIBUTING, CODE_OF_CONDUCT, devcontainer
  5. Release and Maintenance (15%) - releases, CHANGELOG, CI badges, dependabot, recency
  6. Search Discoverability (10%) - keyword placement, GitHub Explore signals, AI citability
CategoryWeightWhat It Checks
README Quality25%Structure, headings, badges, table of contents, code examples
Metadata and Discovery20%Description keywords, topics, homepage URL, feature toggles
Legal Compliance15%License file, SECURITY.md, CITATION.cff, fork obligations
Community Health15%Issue templates, CONTRIBUTING, CODE_OF_CONDUCT, devcontainer
Release and Maintenance15%Releases, CHANGELOG, CI badges, dependabot, recency
Search and Discoverability10%Keyword placement, GitHub Explore signals, AI citability
// Installation

5-Minute Setup

The installer copies all skills and configures DataForSEO + KIE.ai for live keyword data and AI-generated banner images.

claude-github/install.sh
Claude GitHub installer showing splash screen, skill installation, DataForSEO and KIE.ai setup, and available commands
// Standard Operating Procedure

Follow the SOP

The assessment generates a numbered remediation plan. Run each skill in order - each one hands off to the next. Read the full walkthrough for details on each category.

  1. Step 0: /github audit - Diagnose by scoring 6 categories and generating your SOP
  2. Step 1: /github legal - Foundation with license, compliance, and fork obligations
  3. Step 2: /github community - Infrastructure with templates, CoC, and devcontainer
  4. Step 3: /github release - Versioning with CHANGELOG, badges, and catch-up releases
  5. Step 4: /github seo - Research keyword data for description and README
  6. Step 5: /github meta - Settings for description, topics, and features using keyword data
  7. Step 6: /github readme - Capstone README optimization using everything above
  8. Step 7: /github audit - Measure by re-running the review to verify improvement
Step 0
/github audit
Diagnose: scores 6 categories, generates your SOP
Step 1
/github legal
Foundation: license, compliance, fork obligations
Step 2
/github community
Infrastructure: templates, CoC, devcontainer
Step 3
/github release
Versioning: CHANGELOG, badges, catch-up releases
Step 4
/github seo
Research: keyword data for description and README
Step 5
/github meta
Settings: description, topics, features (uses keyword data)
Step 6
/github readme
Capstone: README optimization (uses everything above)
Step 7
/github audit
Measure: re-run the review to verify improvement
// FAQ

Questions

Common Questions About Claude GitHub

Claude Code skills are markdown instruction files (SKILL.md) that extend Claude Code with specialized capabilities. They follow the Agent Skills open standard, meaning any SKILL.md file placed in ~/.claude/skills/ is automatically loaded when Claude Code starts. Each skill defines a trigger pattern (like /github audit), reference files it needs, and sub-agents it can spawn. For example, the Claude GitHub assessment skill triggers on "/github audit", loads 9 reference guides covering README scoring rubrics, legal compliance checklists, and search optimization best practices, then spawns 6 parallel scoring agents. Skills are plain markdown with YAML frontmatter, so they are easy to read, edit, and version control alongside your code. No compilation or build step is required.
Run the installer from your terminal: bash install.sh on macOS/Linux or .\install.ps1 on Windows. The installer copies all 8 skill files to ~/.claude/skills/github/, sets up the DataForSEO MCP server configuration in ~/.claude/.mcp.json, and optionally configures KIE.ai for AI-generated banner images. After installation, restart Claude Code and the skills are available immediately. Type /github to see all available commands. The installer also validates your environment, checking for Claude Code CLI, GitHub CLI (gh), and Node.js. If DataForSEO credentials are not provided during installation, all skills still work using fallback analysis from the GitHub API and built-in reference guides, though keyword recommendations will be marked as unverified.
No. Every skill works without DataForSEO by falling back to codebase analysis, GitHub API data, and 9 built-in reference guides. The /github audit, /github legal, /github community, /github release, /github readme, /github meta, and /github empire commands all function independently. However, keyword recommendations from /github seo will be marked "unverified" without live search data. DataForSEO adds real monthly search volume numbers, keyword difficulty scores, SERP competitor analysis, and search intent classification. A typical single-repository review costs about 15-30 cents in API credits. Portfolio assessments with /github audit username cost proportionally more depending on the number of projects analyzed.
Skills are instruction files (SKILL.md) that Claude loads based on trigger patterns in your message, like "/github audit" or "/github readme". They define what to do, which reference files to consult, and what agents to spawn. Agents are independent sub-processes that skills dispatch to run tasks in parallel. Each agent gets its own context window and can use tools like file reading, web fetching, and code analysis. In Claude GitHub, the /github audit skill spawns 6 scoring agents simultaneously: one each for README Quality, Metadata and Discovery, Legal Compliance, Community Health, Release and Maintenance, and Search Discoverability. They all run concurrently, so a full review completes in under 2 minutes instead of running each check sequentially.
Run /github audit and 6 specialized agents score your repository in parallel. Each agent evaluates one category using a detailed rubric with specific point values, not subjective impressions. The categories are: README Quality (25%) checks structure, headings, badges, code examples, and table of contents. Metadata and Discovery (20%) evaluates description keywords, topics, homepage URL, and feature toggles. Legal Compliance (15%) verifies license files, SECURITY.md, CITATION.cff, and fork obligations. Community Health (15%) checks issue templates, CONTRIBUTING.md, and CODE_OF_CONDUCT.md. Release and Maintenance (15%) looks at CHANGELOG, CI badges, and semantic versioning. Search Discoverability (10%) analyzes keyword placement and AI citability. The final score is a weighted sum of all six categories.
Yes. Run /github audit username to quick-scan all public repositories under that GitHub account. The skill fetches metadata via the GitHub API, ranks projects by stars, forks, and recent activity, then selects the top candidates for deep analysis. Each selected project gets a full 6-agent parallel review. The output is a cross-portfolio report highlighting shared patterns (like missing licenses across multiple codebases) and a prioritized remediation plan. After fixing individual projects, use /github empire for profile-level optimization including generating a profile README, setting up cross-linking between repositories, synchronizing topics, and creating an AI-generated avatar that matches your branding.
// Real Results

Before and After: claude-github Repository

The full optimization workflow ran on the claude-github repository itself. Here is what changed after one pass through the standard operating procedure.

Score Comparison: avalonreset/claude-github

Before Optimization

README Quality12/25
Metadata8/20
Legal3/15
Community2/15
Releases4/15
Discoverability2/10
Total31/100

After One Pass

README Quality23/25
Metadata18/20
Legal14/15
Community13/15
Releases13/15
Discoverability9/10
Total90/100

What Changed

  • README: Added table of contents, badge row, structured headings, and code examples for every command
  • Metadata: Optimized description with target keywords, added 12 topics, set homepage URL, enabled all feature toggles
  • Legal: Generated MIT LICENSE, SECURITY.md with vulnerability reporting policy, and CITATION.cff
  • Community: Created 3 issue templates, CONTRIBUTING.md, CODE_OF_CONDUCT.md, and devcontainer config
  • Releases: Tagged v1.0.0 with semantic versioning, generated CHANGELOG.md, added CI badge
  • Discoverability: Placed keywords in first 100 characters of description, added alt text to images, improved AI citability

Scores reflect the claude-github repository's own assessment run. Results vary by project - repositories with existing documentation typically start higher. The full process took about 15 minutes of developer time plus under 2 minutes of agent processing.

STOP BEING INVISIBLE.

8 skills. 6 parallel agents. Live keyword data. One install command. Your projects deserve to be found. Read our guides on GitHub search keywords and README optimization to learn more.

Copy Install Command GitHub >
// Built By

Meet the Developer

Benjamin, developer of Claude GitHub
Benjamin
Founder, Avalon Reset
Developer and search optimization practitioner building AI-powered tools for GitHub repository visibility. Creator of Claude GitHub and Rankenstein. Focused on making repositories discoverable through structured data, keyword research, and community health standards.
Website GitHub Community Email

About Claude GitHub

Claude GitHub is a repository optimization suite built by Avalon Reset for the Claude Code command-line interface. It is not a product of Anthropic or GitHub, Inc. The suite provides 8 specialized skills and 6 parallel scoring agents that assess, score, and improve GitHub projects across README quality, metadata and discovery, legal compliance, community health, release management, and search visibility. Every recommendation is data-backed through DataForSEO integration for live keyword research with real monthly search volume, keyword difficulty, and SERP competitor analysis. KIE.ai integration provides AI-generated banner images for README files. The suite is open-source under the MIT license and works on Windows, macOS, and Linux. It was scaffolded with skill-forge and uses search optimization methodology adapted from Claude SEO. Built by Avalon Reset.