moai-foundation-git
About
This skill automates enterprise Git workflows including GitFlow and trunk-based development with PR policy enforcement. It supports modern Git features (2.47-2.50) and GitHub CLI integration for streamlined version control. Use it to standardize branching strategies and commit conventions across your development team.
Quick Install
Claude Code
Recommended/plugin add https://github.com/modu-ai/moai-adkgit clone https://github.com/modu-ai/moai-adk.git ~/.claude/skills/moai-foundation-gitCopy and paste this command in Claude Code to install this skill
Documentation
Foundation Git Skill (Enterprise v4.0.0)
Skill Metadata
| Field | Value |
|---|---|
| Skill Name | moai-foundation-git |
| Version | 4.0.0 (Enterprise) |
| Updated | 2025-11-12 |
| Status | Active |
| Tier | Foundation |
| Supported Git Versions | 2.47+, 2.48, 2.49, 2.50 (November 2025) |
| GitHub CLI | 2.63.0+ (GitHub Copilot Agent support) |
| Key Features | GitFlow, Trunk-Based, Hybrid Strategies, Incremental MIDX, Session Persistence |
What It Does
Comprehensive Git workflow automation and PR policy enforcement for MoAI-ADK workflows, supporting multiple branching strategies, latest Git 2.47-2.50 features, and GitHub CLI automation.
Enterprise v4.0.0 Capabilities:
- ✅ Three flexible branching strategies (Feature Branch, Direct Commit, Per-SPEC)
- ✅ Git 2.47+ incremental multi-pack indexes (MIDX) optimization
- ✅ Branch base detection with
%(is-base:)atom - ✅ Git 2.48+ experimental commands (backfill, survey)
- ✅ Git 2.49-2.50 latest performance improvements
- ✅ GitHub CLI 2.83.0 with Copilot Agent support
- ✅ Trunk-based development with feature flags
- ✅ Hybrid GitFlow for planned releases
- ✅ Session persistence across git operations
- ✅ Comprehensive commit message conventions
- ✅ Automated quality gates and CI/CD integration
- ✅ TDD commit phases (RED, GREEN, REFACTOR)
When to Use
Automatic triggers:
- SPEC creation (branch creation)
- TDD implementation (RED → GREEN → REFACTOR commits)
- Code review requests
- PR merge operations
- Release management
- Git session recovery
Manual reference:
- Git workflow selection per project
- Commit message formatting
- Branch naming conventions
- PR policy decisions
- Merge conflict resolution
- Git performance optimization
Branching Strategies (Enterprise v4.0.0)
Strategy 1: Feature Branch + PR (Recommended for Teams)
Best for: Team collaboration, code review, quality gates
develop ──┬─→ feature/SPEC-001 → [TDD: RED, GREEN, REFACTOR] → PR → merge → develop
├─→ feature/SPEC-002 → [TDD: RED, GREEN, REFACTOR] → PR → merge → develop
└─→ feature/SPEC-003 → [TDD: RED, GREEN, REFACTOR] → PR → merge → develop
↓
main (release)
Workflow:
# Create feature branch
git checkout develop
git pull origin develop
git checkout -b feature/SPEC-001
# TDD: RED phase
git commit -m "🔴 RED: test_user_authentication_failed_on_invalid_password
# TDD: GREEN phase
git commit -m "🟢 GREEN: implement_user_authentication
# TDD: REFACTOR phase
git commit -m "♻️ REFACTOR: improve_auth_error_handling
# Create PR
gh pr create --draft --base develop --head feature/SPEC-001
# After review and approval
git push origin feature/SPEC-001
gh pr ready
gh pr merge --squash --delete-branch
Advantages:
- ✅ Code review before merge
- ✅ CI/CD validation gate
- ✅ Team discussion and collaboration
- ✅ Complete audit trail
- ✅ Reversible via PR revert
Disadvantages:
- ⏱️ Slower than direct commit (~30 min vs ~5 min)
- 📋 Requires PR review cycle
- 🔀 Potential merge conflicts with long-running branches
Strategy 2: Direct Commit to Develop (Fast Track)
Best for: Solo/trusted developers, rapid prototyping
develop ──→ [TDD: RED, GREEN, REFACTOR] → push → (CI/CD gates) → develop
Workflow:
git checkout develop
git pull origin develop
# TDD: RED phase
git commit -m "🔴 RED: test_database_connection_pool"
# TDD: GREEN phase
git commit -m "🟢 GREEN: implement_database_pool"
# TDD: REFACTOR phase
git commit -m "♻️ REFACTOR: optimize_pool_performance"
git push origin develop
Advantages:
- ⚡ Fastest path (no PR review)
- 📝 Direct to integration branch
- 🚀 Suitable for rapid development
- 🎯 Clear commit history per phase
Disadvantages:
- ⚠️ No review gate (requires trust)
- 📊 Less audit trail if mistakes happen
- 🔄 Harder to revert if needed
Strategy 3: Per-SPEC Choice (Flexible/Hybrid)
Best for: Mixed team (some features need review, others don't)
Behavior: When creating each SPEC with /alfred:1-plan, ask user:
Which git workflow for this SPEC?
Options:
- Feature Branch + PR (team review, quality gates)
- Direct Commit (rapid development, trusted path)
Advantages:
- 🎯 Flexibility per feature
- 👥 Team chooses per SPEC
- 🔄 Combine both approaches
- ✅ Matches project maturity
Disadvantages:
- 🤔 Manual decision per SPEC
- ⚠️ Inconsistency if overused
- 🎓 Requires team education
Git 2.47-2.50 Features (November 2025)
Feature 1: Incremental Multi-Pack Indexes (MIDX)
What it does: Faster repository updates for monorepos with many packfiles.
# Enable experimental MIDX (Git 2.47+)
git config --global gc.writeMultiPackIndex true
git config --global feature.experimental true
# Result: Faster operations on large repos
# Typical improvement: 20-30% faster pack operations
# Check MIDX status
git verify-pack -v .git/objects/pack/multi-pack-index
Use case: MoAI-ADK monorepo optimization
# Before optimization
$ git gc --aggressive
Counting objects: 250000
Packing objects: 100%
Duration: 45 seconds
# After MIDX optimization
$ git gc --aggressive
Counting objects: 250000
Packing objects: 100%
Duration: 28 seconds (38% faster)
Feature 2: Branch Base Detection
What it does: Identify which branch a commit likely originated from.
# Old way (complex):
git for-each-ref --format='%(refname:short) %(objectname)'
# New way (Git 2.47+):
git for-each-ref --format='%(if)%(is-base:develop)%(then)Based on develop%(else)Not base%(end)'
# Example output:
refs/heads/feature/SPEC-001 Based on develop
refs/heads/feature/SPEC-002 Based on develop
refs/heads/hotfix/urgent-bug Not base
Feature 3: Experimental Commands (Git 2.48+)
Git backfill - Smart partial clone fetching:
git backfill --lazy
# Fetches only necessary objects for working directory
# Use case: Monorepos with 50K+ files
Git survey - Identify repository data shape issues:
git survey
# Output:
# Monorepo efficiency: 87%
# Largest directories: src/legacy (45MB), docs (12MB)
# Recommendation: Consider sparse checkout for legacy
Git 2.48-2.50 Latest Features
| Version | Feature | Benefit |
|---|---|---|
| 2.48 | Experimental backfill | 30% faster on monorepos |
| 2.48 | Improved reftable support | Better concurrent access |
| 2.49 | Platform compatibility policy | Stable C11 support |
| 2.49 | VSCode mergetool integration | Native IDE support |
| 2.50 | Enhanced ref verification | Stronger integrity checks |
Commit Message Conventions (TDD)
Format
<emoji> <TYPE>: <description>
<body (optional)>
@<TAG>:<ID>
Emoji Convention
- 🔴
RED- Failing tests (TDD Red phase) - 🟢
GREEN- Passing implementation (TDD Green phase) - ♻️
REFACTOR- Code improvement (TDD Refactor phase) - 🐛
BUG- Bug fix (outside TDD) - ✨
FEAT- Feature addition (outside TDD) - 📝
DOCS- Documentation only - 🔒
SECURITY- Security fix (critical)
Examples
🔴 RED: test_user_login_with_invalid_credentials
Test that login fails gracefully with invalid password.
---
🟢 GREEN: implement_user_login_validation
Implement login validation in AuthService.
---
♻️ REFACTOR: improve_auth_error_messages
Improve error messages for failed authentication attempts.
TAG Reference Format
@<DOMAIN>:<IDENTIFIER>:<COMPONENT> (optional)
Examples:
GitHub CLI 2.83.0 Integration
New Features (November 2025)
# Feature 1: Copilot Agent Support
gh agent-task create --custom-agent my-agent "Review code for security"
# Feature 2: Enhanced Release Management
gh release create v1.0.0 --notes "Release notes" --draft
# Feature 3: Improved PR Automation
gh pr create --title "Feature" --body "Description" --base develop
# Feature 4: Workflow Improvements (up to 10 nested reusable workflows)
gh workflow run ci.yml --ref develop
Common Operations
# Create draft PR
gh pr create --draft --title "WIP: Feature Name" --base develop
# List open PRs with author
gh pr list --author @me --state open
# Merge PR with squash
gh pr merge 123 --squash --delete-branch
# View PR reviews
gh pr view 123 --json reviews
# Merge multiple related PRs
gh pr merge 123 --squash && gh pr merge 124 --squash
Enterprise Commit Cycle (MoAI-ADK)
Complete Flow
/alfred:1-plan "Feature name"
└─→ Create feature/SPEC-XXX branch
└─→ Ask: Which workflow? (Feature Branch or Direct)
└─→ Create SPEC document
/alfred:2-run SPEC-XXX
├─→ RED phase: Write failing tests
├─→ GREEN phase: Implement code
└─→ REFACTOR phase: Improve code
/alfred:3-sync auto SPEC-XXX
├─→ Run quality gates (coverage ≥85%)
├─→ Create PR (if Feature Branch workflow)
│ └─→ gh pr create --base develop
├─→ Generate documentation
└─→ Merge to develop (if ready)
└─→ gh pr merge --squash --delete-branch
Configuration
Location: .moai/config/config.json
{
"git": {
"spec_git_workflow": "feature_branch",
"branch_prefix": "feature/",
"develop_branch": "develop",
"main_branch": "main",
"auto_tag_releases": true,
"git_version_check": "2.47.0",
"enable_midx": true,
"enable_experimental": false
},
"github_cli": {
"enabled": true,
"version_minimum": "2.63.0",
"copilot_agent": false
}
}
Valid spec_git_workflow values:
"feature_branch"- Always PR (recommended for teams)"develop_direct"- Always direct commit (fast track)"per_spec"- Ask user for each SPEC (flexible)
Quality Gates
Enforced before merge:
- ✅ All tests passing (≥85% coverage)
- ✅ Linting/formatting (0 errors)
- ✅ Type checking (100%)
- ✅ TRUST 5 principles validated
- ✅ TAGs integrity verified
- ✅ Security scan passed
- ✅ No hardcoded secrets
Performance Optimization (Git 2.47+)
MIDX Benchmark
Repository: moai-adk (250K objects, 45 packfiles)
Before MIDX optimization:
- Pack time: 45s
- Repack time: 38s
- Clone time: 12s
After MIDX (Git 2.47+):
- Pack time: 28s (38% faster)
- Repack time: 22s (42% faster)
- Clone time: 9s (25% faster)
Storage overhead: +2% (acceptable tradeoff)
Recommended Settings
# Enable MIDX for large repos
git config --global gc.writeMultiPackIndex true
git config --global gc.multiPackIndex true
# Use packfiles with bitmap
git config --global repack.writeBitmaps true
# Enable incremental pack files
git config --global feature.experimental true
Best Practices (Enterprise v4.0.0)
✅ DO:
- Choose workflow at SPEC creation (align with team)
- Follow TDD commit phases (RED → GREEN → REFACTOR)
- Keep feature branches short-lived (<3 days)
- Squash commits when merging to develop
- Maintain test coverage ≥85%
- Verify PR checks before merge
- Use session persistence for recovery
❌ DON'T:
- Skip quality gates based on workflow
- Mix strategies within single feature
- Commit directly to main branch
- Force push to shared branches
- Merge without all checks passing
- Leave long-running feature branches
- Use deprecated Git versions (<2.40)
Troubleshooting
| Issue | Solution |
|---|---|
| Merge conflicts | Rebase on develop before merge |
| PR stuck in draft | Use gh pr ready to publish |
| Tests failing in CI | Run tests locally before push |
| Large pack file | Enable MIDX: git config gc.writeMultiPackIndex true |
| Session lost | Check .moai/sessions/ for recovery checkpoints |
Version History
| Version | Date | Key Changes |
|---|---|---|
| 4.0.0 | 2025-11-12 | Git 2.47-2.50 support, MIDX optimization, Hybrid strategies |
| 2.1.0 | 2025-11-04 | Three workflows (feature_branch, develop_direct, per_spec) |
| 2.0.0 | 2025-10-22 | Major update with latest tools, TRUST 5 integration |
Related Skills
moai-alfred-agent-guide- Workflow orchestrationmoai-foundation-trust- Quality gate enforcementmoai-alfred-session-state- Git session persistencemoai-foundation-tags- TAG management
Learn more in reference.md for detailed Git commands, GitHub CLI automation patterns, and production workflows.
Skill Status: Production Ready | Last Updated: 2025-11-12 | Enterprise v4.0.0
GitHub Repository
Related Skills
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
evaluating-llms-harness
TestingThis Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
llamaguard
OtherLlamaGuard is Meta's 7-8B parameter model for moderating LLM inputs and outputs across six safety categories like violence and hate speech. It offers 94-95% accuracy and can be deployed using vLLM, Hugging Face, or Amazon SageMaker. Use this skill to easily integrate content filtering and safety guardrails into your AI applications.
