GitAI - AI-Native Version Control System

Traditional code review: Check code security and efficiency AI-era code review: Check prompt intent, token costs, and ROI

Git Compatible · Lightweight CLI · Pure Local · 100% Go

---

🎯 Philosophy

The Shift to AI-Native Development

The AI era has fundamentally changed software development:

Aspect	Traditional Era	AI Era
What we write	Code	Prompts
What we review	Security, Performance	Intent, Cost, Effectiveness
What we measure	Lines of Code	Token Efficiency, ROI
Version control	Git manages code	What manages prompts?

GitAI fills this gap - it's an AI-native version control system that treats prompts as first-class citizens alongside your code.

Core Principles

1. Intent as First-Class Data

   • Every commit tracks: why it was made, what prompts were used, how much it cost
   • Transform commits from opaque code changes into traceable intent chains

2. Prompts Are Code, But Different

   • Prompts have versions, dependencies, and quality metrics
   • Token consumption is a real cost that needs tracking
   • Prompt changes can have rippling effects across codebases

3. Bidirectional Traceability

   • Forward: Intent → Prompts → Code → Behavior
   • Backward: Code → Prompts → Intents → Decisions
   • Nothing gets lost in the AI development loop

4. Continuous Learning

   • Track what works and what doesn't
   • Build institutional knowledge from feedback
   • Optimize prompts based on real outcomes

---

✨ Key Features

📋 Prompt Version Control

• Track .prompt files alongside code
• View token changes between versions (gitai diff)
• See commit history with token deltas (gitai log)
• Structured prompt format: role, task, constraints, examples, input, output

💰 Cost Intelligence

• Calculate costs per file, section, role, user, or branch
• Multi-model support: OpenAI, Qwen, Wenxin, and custom
• Currency conversion: USD/CNY with configurable exchange rates
• Visualize spending trends over time (gitai trend)

🔍 Impact Analysis

• See what changes if you modify a prompt (gitai impact <prompt>)
• Auto-detect prompt→code dependencies (gitai scan-deps)
• Trace which prompts affect which files
• Find prompts that share affected files

🧪 Quality Assurance

• Automated prompt quality scoring (clarity, conciseness, efficiency)
• Pre-commit hooks to catch low-quality prompts
• Review suggestions for improvement
• Quality dashboard (gitai review, gitai quality)

🎯 Intent Tracking

• Every commit records developer intent
• Automatic category detection (fix/feature/refactor/ai-gen/experimental)
• Link commits to the prompts that generated them
• Full intent history (gitai intent-log)

📚 Organization

• Templates: Create reusable prompt templates (gitai template)
• Tags: Organize prompts with custom tags (gitai tag)
• Search: Full-text search across all prompts
• Export/Import: Backup and restore prompt library

🧠 Feedback Loop

• Record outcomes: good/bad/neutral feedback (gitai feedback)
• Track satisfaction metrics over time
• Learn from results to optimize prompts
• AI-generated insights from your data (gitai insights)

🔗 Full Traceability

• Trace intent → prompts → affected files (gitai trace <intent>)
• Trace file → prompts → intents (gitai trace <file>)
• See the complete chain from idea to implementation
• Understand downstream effects of changes

⚙️ Automation

• Git hooks for automatic quality checks
• Pre-commit: validate prompt quality before committing
• Post-commit: automatically record intent
• No manual overhead - it just works

---

🚀 Quick Start

# 1. Initialize in your Git repository
git init
gitai init

# 2. Track your first prompt
gitai track prompts/agent.prompt

# 3. See what changed
gitai diff                    # Token changes
gitai log                     # Commit history
gitai cost --user             # Cost by developer
gitai impact prompts/agent.prompt  # Impact analysis

# 4. Learn from your data
gitai insights                # AI-generated insights
gitai trace <intent-id>       # Full trace chain

---

📖 Command Reference

Getting Started

Command	Description
gitai init	Initialize .gitai/ in current repository
gitai track <file>	Track a .prompt file
gitai status	Show tracked files status
gitai show <file>	Show detailed file information

Version Control

Command	Description
gitai diff <file>	Compare prompt versions with token changes
gitai log	Show commit history with token deltas
gitai commit	Create commit with intent tracking

Cost Analysis

Command	Description
gitai cost	Show cost breakdown (by file/section/role)
gitai cost --user	Cost breakdown by developer
gitai cost --branch	Cost breakdown by branch
gitai price	List model pricing
gitai stats	Comprehensive statistics dashboard
gitai trend	Token growth over time

Quality & Review

Command	Description
gitai review	Quality analysis with improvement suggestions
gitai quality	Quick quality overview
gitai hooks install	Install automatic quality checks

Organization

Command	Description
gitai template create	Create prompt template
gitai template use	Use template to create file
gitai tag add <file> <tag>	Add tags to file
gitai search <query>	Search across prompts
gitai export	Export to JSON/CSV
gitai import	Import from export

AI-Era Features

Command	Description
gitai impact <prompt>	Analyze impact of changing a prompt
gitai scan-deps	Auto-detect prompt dependencies
gitai show-deps	Show dependency graph
gitai feedback [good|bad]	Record feedback on commits
gitai insights	Show AI-generated insights
gitai trace <id|file>	Trace intent↔code relationships

---

💡 Use Cases

1. Understand Your AI Spending

$ gitai cost --user
Cost breakdown by user (model: gpt-4o)

User          Commits  Files  Tokens    USD      CNY
----          -------  -----  ------    ---      ---
alice         15       23     45,230    $0.12    ¥0.86
bob           8        12     12,400    $0.03    ¥0.22

TOTAL         23       35     57,630    $0.15    ¥1.08

2. Trace an Idea to Implementation

$ gitai trace intent-abc123456
🔍 Trace: intent-abc123456

Commit: abc1234
Message: feat: add user authentication

📝 Prompts Used (2):
  → prompts/auth-api.prompt
  → prompts/security-check.prompt

📁 Affected Files (5):
  → api/auth.go
  → api/middleware.go
  → models/user.go
  → utils/crypto.go
  → tests/auth_test.go

3. Analyze Impact Before Changes

$ gitai impact prompts/api-design.prompt
📊 Impact Analysis: prompts/api-design.prompt

Affected Files: 3 files
  → api/handlers.go
  → api/routes.go
  → models/types.go

⚠️  Related Prompts (2)
These prompts share affected files:
  • prompts/api-docs.prompt
  • prompts/api-tests.prompt

4. Learn From Your Data

$ gitai insights
💡 GitAI Insights

Total Commits Tracked:  47
Total Intents:           42
Total Feedback:          15
Average Score:           78.5/100

📊 Activity by Category:
  feature: 18
  fix: 12
  refactor: 8
  ai-gen: 4

✨ Key Learnings:
  • Most active category: feature (18 intents)
  • High satisfaction rate - prompts are working well
  • Consider standardizing common patterns

🔝 Most Used Prompts:
  1. prompts/api-helper.prompt (12 uses)
  2. prompts/code-reviewer.prompt (8 uses)
  3. prompts/test-generator.prompt (5 uses)

---

📦 Installation

From Source

git clone https://github.com/YuanyuanMa03/gitai.git
cd gitai
go build -o gitai ./cmd/gitai
sudo mv gitai /usr/local/bin/

Using Go Install

go install github.com/YuanyuanMa03/gitai/cmd/gitai@latest

---

🏗️ Architecture

gitai/
├── cmd/gitai/              # CLI entry point
├── internal/gitai/         # Core commands
│   ├── git.go              # Git integration
│   ├── storage.go          # Storage layer
│   ├── intent.go           # Intent tracking
│   ├── feedback.go          # Feedback system
│   ├── dependency.go       # Dependency graph
│   ├── hooks.go            # Git hooks
│   ├── insights.go         # Analytics & trace
│   └── ...                 # Other commands
├── pkg/token/              # Token calculation
│   ├── estimator.go        # Pure Go estimation
│   └── parser.go           # Prompt parser
└── go.mod

Technology Stack

Component	Technology
Language	Go 1.21+
Git Integration	go-git
CLI Framework	urfave/cli/v2
Storage	JSON (.gitai/*.json)
Configuration	TOML (.gitai/config.toml)

Design Principles

1. No Database - JSON file storage, simple and portable
2. No Web UI - Pure CLI, fast and scriptable
3. No Network - 100% local, private and fast
4. Git Compatible - Works with any existing Git repo
5. Minimal Dependencies - Easy to maintain and audit

---

📊 Token Estimation

GitAI uses pure Go implementation for accurate token estimation:

Character Type	Token Ratio
Chinese (CJK)	~0.7 tokens/char
English letters	~0.25 tokens/char
Numbers	~0.3 tokens/char
Punctuation	~0.5 tokens/char

Why local estimation?

• ✅ Zero network latency
• ✅ Zero API costs
• ✅ Works offline
• ✅ Privacy-preserving
• ✅ Sufficient for cost planning (~10% margin)

---

🧪 Testing

# Run all tests
go test ./...

# Run with coverage
go test -cover ./...

# Run specific test
go test -v ./internal/gitai -run TestIntent

GitAI is built with Test-Driven Development (TDD) - all features have comprehensive test coverage.

---

📝 License

MIT License - see LICENSE for details.

---

🤝 Contributing

Contributions welcome! Please feel free to submit issues and pull requests.

Development Setup

git clone https://github.com/YuanyuanMa03/gitai.git
cd gitai
go mod tidy
go build -o gitai ./cmd/gitai
./gitai --help

---

📮 Contact

• GitHub: @YuanyuanMa03
• Issues: GitHub Issues

---

🙏 Acknowledgments

• go-git - Git pure Go implementation
• urfave/cli - Command-line interface framework
• BurntSushi/toml - TOML parser for Go

---

GitAI - Where prompts meet version control.