π€ Your AI-Powered Development Assistant - Works Locally & On GitHub
SuperAI GitHub provides 10 specialized AI agents, tech-specific skills, and GitHub Actions workflows that work seamlessly with Claude Code. Install as a submodule in any repository to get instant AI assistance.
- π€ 10 Specialized Agents - From code review to security audits
- π Tech-Specific Skills - React, Next.js, Laravel, Python, Docker, K8s, and more
- π§ GitHub Integration - Automatic PR reviews, security scans, issue triage
- π― Smart Detection - Auto-detects your tech stack and installs relevant agents
- π Prompt Templates - Reusable templates for common tasks
- π Non-Invasive - Never touches your
.claude/directory
# 1. Add as submodule
git submodule add https://github.com/enact-on/super-ai-github.git .super-ai-github
cd .super-ai-github
# 2. Run installer
./scripts/install.sh
# 3. Use Claude Code normally
claude-code.claude-agents/- All agents and skills (NEVER touches your.claude/).github/workflows/- Automated workflows for GitHub Actions- Tech-specific agents - Based on your project's detected tech stack
All agents are in .claude-agents/agents/ as markdown files:
# Use the orchestrator to delegate tasks
"Read the orchestrator.md file and help me implement this feature"
# Use specific specialists
"Read the security-auditor.md file and audit this code for security issues"
"Read the frontend-specialist.md file and review this React component"| Agent | Role | Best For |
|---|---|---|
| Orchestrator | Team lead | Complex task coordination |
| Code Reviewer | Quality analysis | PR reviews, code quality |
| Security Auditor | Security analysis | Security audits, vulnerability scanning |
| Implementation | Code writing | Feature implementation |
| Frontend Specialist | React/Vue/UI | Frontend development |
| Backend Specialist | API/Server logic | Backend development |
| Database Expert | DB design | Database optimization |
| Testing Specialist | Test strategies | Test implementation |
| DevOps Specialist | Docker/K8s/CI/CD | Infrastructure |
| Documentation Writer | Technical docs | Documentation |
# React projects
"Read .claude-agents/skills/react/SKILL.md and apply these patterns"
# Next.js projects
"Read .claude-agents/skills/nextjs/SKILL.md and help with Server Components"
# Testing
"Read .claude-agents/skills/testing/SKILL.md and write tests for this code"-
Add GitHub Secret:
- Go to repository Settings β Secrets and variables β Actions
- Add:
ANTHROPIC_API_KEY(your Anthropic API key)
-
Push to GitHub:
git add . git commit -m "Add SuperAI GitHub integration" git push
- π PR Reviews - Every PR gets automatic code review
- π Security Audits - Weekly security scans (Sundays at midnight UTC)
- π·οΈ Issue Triage - Automatic issue classification and labeling
- π¬ Manual Commands - Use
/superai helpon any issue/PR
# Code review
"Read .claude-agents/prompts/code-review-prompt.md and review this PR"
# Security audit
"Read .claude-agents/prompts/security-audit-prompt.md and audit this code"
# Implementation
"Read .claude-agents/prompts/implementation-prompt.md and implement this feature"your-project/
βββ .claude/ # Your existing config (NOT touched) β
βββ .claude-agents/ # SuperAI agents and skills (NEW) π
β βββ agents/ # 10 specialized agents
β βββ skills/ # Tech-specific skills
β βββ prompts/ # Reusable prompts
β βββ settings.json # Configuration
βββ .github/workflows/ # Automated workflows (NEW) π
βββ .super-ai-github/ # Submodule reference
# Start Claude Code
claude-code
# Ask it to use agents
"Read the orchestrator.md file and coordinate a code review for this PR"
"Read the testing-specialist.md file and write tests for this function"
"Read .claude-agents/skills/react/SKILL.md and refactor this component"# Automatic (just create PR)
git checkout -b feature-branch
# Make changes
git push origin feature-branch
# Create PR on GitHub β Automatic review happens
# Manual (on any issue/PR)
/superai help
/superai review this code
/superai explain the authentication logicThe installer automatically detects your tech stack:
package.json β React, Next.js, Vue, Node.js, TypeScript
composer.json β Laravel, PHP
requirements.txt β Python, Django, FastAPI
go.mod β Go# Install all agents and skills
./scripts/install.sh --allcd .super-ai-github
git pull origin main
./scripts/install.sh- USAGE.md - Comprehensive usage guide
- docs/GITHUB_SETUP.md - Detailed GitHub setup
- MIGRATION_SUMMARY.md - Migration overview
- Install - Run the installer script
- Reference - Point Claude Code to agent/skill files
- Execute - Agents provide specialized assistance
- Install - Run the installer script
- Push - Push to GitHub with
ANTHROPIC_API_KEYsecret - Automate - Workflows run automatically on PRs, issues, schedules
# Create your own agent
echo "# My Custom Agent
Custom instructions here..." > .claude-agents/agents/my-agent.md# Create your own skill
mkdir -p .claude-agents/skills/my-skill
echo "# My Skill
Custom patterns here..." > .claude-agents/skills/my-skill/SKILL.mdUser Request
β
Orchestrator (coordinates)
β
Technology Detection + Skill Loading
β
Specialist Agent (executes)
β
Code Reviewer | Security | Implementation | etc.
β
Results (quality assistance)
- Instant AI Assistance - No setup required, just reference agent files
- Specialized Expertise - Each agent focuses on their domain
- Consistent Quality - Same standards across all projects
- Learning Opportunity - Get feedback from AI specialists
- Scalable Code Review - Handle 100+ repositories easily
- Automated Security - Regular vulnerability scans
- Standardized Processes - Same review criteria everywhere
- Reduced Manual Work - Automate routine tasks
- Cost Effective - One API key, unlimited usage
- Easy Onboarding - Simple install for new repos
- Flexible Deployment - Works with any tech stack
- Centralized Updates - Update once, affect all repos
- Never touches
.claude/- Your local config is safe - API Key in GitHub Secrets - Secure credential storage
- Read-only agents - Review agents can't modify code
- Standard Anthropic API - No custom providers required
Contributions welcome! See docs/CONTRIBUTING.md for guidelines.
MIT License - See LICENSE file for details
SuperAI GitHub - Your AI-powered development assistant π
Part of the SuperAI ecosystem - Centralized AI development for teams
- 10 Role-based Agents - Orchestrator delegates to specialized agents (code reviewer, security auditor, testing specialist, DevOps specialist, etc.)
- 18 Technology Skills - Reusable patterns for React, Vue, Next.js, Laravel, Node.js, Python, Docker/K8s, and more
- GitHub Integration - Automatic PR reviews, scheduled security scans, issue triage, comment-based triggers
- Claude Code Native - Uses standard Claude Code configuration and workflows
- Zero Configuration - Uses standard
.claude/directory structure - Custom Provider Support - Works with custom Anthropic-compatible providers
-
Clone or add as submodule to your project:
# Option 1: Clone directly git clone https://github.com/enact-on/super-ai-github.git cd super-ai-github ./scripts/install.sh # Option 2: Add as submodule git submodule add https://github.com/enact-on/super-ai-github.git .super-ai-source cd .super-ai-source ./scripts/install.sh
-
Use locally:
cd /path/to/your/project claude-code -
Check available agents: Type
/agentsin Claude Code
- Add
ANTHROPIC_API_KEYto your repository secrets - Add
CUSTOM_ANTHROPIC_BASE_URLif using a custom provider - Push to GitHub
- Open a PR to see SuperAI review automatically
orchestrator (Primary)
βββ @code-reviewer β Code quality analysis (read-only)
βββ @security-auditor β Security analysis with CWE/CVE (read-only)
βββ @implementation β Code implementation
βββ @frontend-specialist β React, Vue, UI/UX
βββ @backend-specialist β API design, server logic
βββ @database-expert β Schema, queries, migrations
βββ @testing-specialist β Unit/Integration/E2E tests
βββ @devops-specialist β Docker, K8s, CI/CD
βββ @documentation-writer β README, API docs, guides
This matrix approach avoids combinatorial explosion (e.g., security-react, security-laravel) while allowing specialized expertise.
super-ai-github/
βββ .claude/ # Claude Code configuration
β βββ settings.json # Main Claude Code settings
βββ .claude-agents/ # Agent definitions
β βββ agents/ # 10 Role-based agents
β β βββ orchestrator.md
β β βββ code-reviewer.md
β β βββ security-auditor.md
β β βββ implementation.md
β β βββ frontend-specialist.md
β β βββ backend-specialist.md
β β βββ database-expert.md
β β βββ testing-specialist.md
β β βββ devops-specialist.md
β β βββ documentation-writer.md
β βββ skills/ # Technology-specific skills
β βββ prompts/ # Reusable prompt templates
βββ scripts/
β βββ install.sh # Main installer
β βββ detect-tech-stack.sh # Tech detection
β βββ update.sh # Update to latest
βββ .github/workflows/
βββ claude-review.yml # PR review (all PRs)
βββ claude-security.yml # Weekly security scan
βββ claude-issues.yml # Issue triage
βββ claude-comment.yml # /superai command
| Agent | Role | Temperature | Tools |
|---|---|---|---|
orchestrator |
Team lead, delegates tasks | 0.3 | Full access |
code-reviewer |
Code quality review | 0.1 | Read-only |
security-auditor |
Security analysis (CWE/CVE) | 0.2 | Read-only |
implementation |
Code implementation | 0.2 | Full access |
frontend-specialist |
Frontend development | 0.3 | Full access |
backend-specialist |
Backend development | 0.3 | Full access |
database-expert |
Database design | 0.2 | Read + limited bash |
testing-specialist |
Testing (unit/integration/E2E) | 0.2 | Full access |
devops-specialist |
Docker, K8s, CI/CD | 0.2 | Full access |
documentation-writer |
Documentation | 0.4 | Read/Write (no bash) |
| Skill | Technology | When to Use |
|---|---|---|
react |
React, Next.js, TypeScript | React ecosystem projects |
vue |
Vue.js 3, Nuxt | Vue ecosystem projects |
nextjs |
Next.js 14+ App Router | Server Components, Server Actions |
typescript |
TypeScript | Advanced type patterns |
| Skill | Technology | When to Use |
|---|---|---|
nodejs |
Node.js, Express | Node.js server development |
nestjs |
NestJS framework | Enterprise Node.js APIs |
laravel |
Laravel, PHP | Laravel PHP projects |
python |
Python, FastAPI, Django | Python backend development |
| Skill | Technology | When to Use |
|---|---|---|
docker-k8s |
Docker, Kubernetes | Containerization, orchestration |
api-design |
REST, GraphQL, OpenAPI | API design patterns |
testing |
Jest, Vitest, PHPUnit, pytest | Testing strategies |
| Skill | Technology | When to Use |
|---|---|---|
documentation |
Documentation standards | Writing documentation |
frontend-design |
UI/UX best practices | Frontend design decisions |
code-review |
Code review patterns | Code reviews |
security-guidance |
Security best practices | Security considerations |
Every PR is automatically reviewed by SuperAI when opened or updated. The review analyzes:
- Code Quality - Readability, maintainability, consistency
- Potential Issues - Edge cases, null checks, resource leaks, performance
- Testing - Coverage, edge cases, test quality
- Documentation - Comments, API changes, breaking changes
- Linked Issues - Verifies PR addresses linked issues
Runs every Sunday at midnight UTC, checking for:
- OWASP Top 10 vulnerabilities with CWE references
- Dependency vulnerabilities with CVE references
- Exposed secrets - API keys, passwords, tokens
- Authentication/authorization issues
- Creates GitHub issues for critical/high findings
New issues are automatically triaged with:
- Duplicate Detection - Searches for related issues
- Issue classification (bug, feature, question, performance)
- Priority assessment (critical, high, medium, low)
- Suggested GitHub labels
- Initial response for common issues
Trigger SuperAI manually on issues/PRs:
/superai <your request>/sai <your request>/claude <your request>
- Claude Code v1.0 or higher
- Node.js 18+ (for installer scripts)
- Bash shell (Linux/macOS/WSL)
- GitHub account (for GitHub integration)
MIT
SuperAI GitHub - Centralized Claude Code configuration for organization-wide AI development
To update to the latest version:
cd .super-ai-source
./scripts/update.shSee CONTRIBUTING.md for guidelines on adding agents and skills.
- Create
.claude-agents/agents/my-agent.mdwith proper structure - Include clear description of the agent's role and expertise
- Update orchestrator.md to include the new agent in delegation logic
- Create
.claude-agents/skills/my-skill/SKILL.md - Include comprehensive patterns, examples, and best practices
- Update orchestrator.md to include the new skill
ANTHROPIC_API_KEY- Your Anthropic API key (or custom provider key)CUSTOM_ANTHROPIC_BASE_URL- Optional: Custom provider base URLCLAUDE_CODE_CONFIG_DIR- Defaults to.claude-agents
For GitHub integration, add these secrets to your repository:
ANTHROPIC_API_KEY- Required: API key for Claude CodeCUSTOM_ANTHROPIC_BASE_URL- Optional: Custom provider URL