gitspect

Gitspect

Git history analysis that reveals what your commit messages won't tell you.

Where you struggled. What keeps breaking. Who owns what. Why your project feels slow.

Your repository remembers everything. Gitspect tells you what matters.

What is Gitspect?

Gitspect transforms commit data into actionable insights about:

• Risk areas - Files that are bug-prone, unstable, or over-engineered
• Ownership - Who owns which files, coupling patterns, bus factor
• Evolution - Growth trends, stability changes, velocity over time
• Team patterns - Activity heatmaps, burnout signals, collaboration health

Privacy-first: 100% offline analysis. No APIs, no cloud, no data leaving your machine.

Built for AI workflows: Auto-generates SKILL.md files that help AI assistants understand your repository.

Installation

npm install -g gitspect

Or run directly with npx:

npx gitspect <command>

Quick Start

# First-time setup (creates config and AI skills)
gitspect init

# Personal retrospective - what did I work on?
gitspect reflect

# Find risky files
gitspect scars

# Comprehensive overview for AI assistants
gitspect context --json

Commands

Setup

Command	Description
gitspect init	Initialize .gitspectrc config and auto-generate skills/gitspect/SKILL.md for AI agents

Example:

gitspect init
# ✓ Created .gitspectrc
# ✓ Created skills/gitspect/SKILL.md

Phase 1: Personal Retrospective

For vibe coders who need context recovery and personal insights.

Command	Description
gitspect reflect	Personal retrospective with activity summary, file churn, and insights
gitspect churn	File churn detection - identify frequently modified files
gitspect heatmap	Activity heatmap - visualize when you code most

Example:

gitspect reflect --days 30

Phase 2: Contributor Onboarding

For understanding codebase ownership and risk areas.

Command	Description
gitspect blame-map	File ownership mapping - who owns which files
gitspect scars	Riskiest files to modify - high-churn, bug-prone areas
gitspect couples	Files changed together - discover hidden dependencies

Examples:

# Show top 10 riskiest files
gitspect scars --limit 10

# Check risk level of a specific file (great for AI context)
gitspect scars --file src/components/Button.tsx --json

Phase 3: Project Management

For understanding project health and blockers.

Command	Description
gitspect report	Project health report - metrics, concerns, and positive signals
gitspect blockers	What's slowing progress - hotspots, instability, complexity
gitspect evolution	Codebase evolution over time - trends and growth patterns

Example:

gitspect blockers --days 90

AI Context

Comprehensive repo overview designed for AI assistants.

Command	Description
gitspect context	AI context - comprehensive repo overview with risk assessment

Example:

gitspect context --json > repo_context.json

Global Options

All commands support these options:

Option	Description
--days <n>	Time period in days (default: all time, all branches)
--current-branch	Only analyze the current branch (default: all branches)
--json	Output as JSON (recommended for AI consumption)
--no-ignore	Include files that would normally be filtered (lock files, build artifacts, etc.)
--limit <n>	Limit output to top N results (churn, scars, couples)
--file <path>	Filter analysis to a specific file (scars command)
--granularity <week|month>	Time granularity for evolution command

Tip for AI workflows: Use the --json flag to get clean, machine-readable output. Without it, the output includes colors, tables, and formatting that AI tools may not parse correctly.

Configuration

Create .gitspectrc in your repository root to customize behavior:

{
  "exclude": [
    "tests/fixtures/",
    "*.mock.ts",
    "docs/"
  ],
  "include": [],
  "skillPrompt": "auto"
}

Options

Setting	Description
exclude	File patterns to ignore during analysis (lock files, build artifacts, etc.)
include	File patterns to explicitly include (overrides exclude)
skillPrompt	Control AI skill file behavior: auto (prompt if needed), always (auto-update), never (don't create)

Built-in Ignores

Gitspect automatically filters out common noise files:

• Lock files: package-lock.json, yarn.lock, pnpm-lock.yaml, etc.
• Build artifacts: dist/, build/, *.min.js, etc.
• Generated files: *.generated.*, *.gql.ts, swagger files, etc.

Use --no-ignore to include all files if needed.

AI Integration

Gitspect is designed to work with AI assistants (Claude, Cursor, ChatGPT, etc.).

SKILL.md Auto-Generation

When you run gitspect init, it creates skills/gitspect/SKILL.md - a context file that AI agents automatically use to understand your repository.

The SKILL.md includes:

• When to use Gitspect commands
• What each command reveals
• How to interpret the output
• Project-specific patterns (can be customized)

Auto-update behavior:

• skillPrompt: "auto" - Prompts to update SKILL.md when Gitspect version changes
• skillPrompt: "always" - Automatically keeps SKILL.md updated
• skillPrompt: "never" - Disables SKILL.md management

For AI Agents

Important: Always use the --json flag when providing output to AI assistants. Without it, the output contains ANSI colors, table formatting, and text wrapping that AI tools cannot parse correctly.

Run gitspect context --json to get a comprehensive repo overview:

{
  "overview": {
    "health": "moderate",
    "totalCommits": 147,
    "activeContributors": 3,
    "primaryLanguage": "ts",
    "developmentVelocity": "high"
  },
  "criticalAreas": {
    "highRiskFiles": [{
      "path": "src/payment.ts",
      "riskScore": 73,
      "why": "3x higher churn than average; 28% bugfix rate",
      "recommendation": "proceed with caution, add tests"
    }]
  },
  "ownership": {
    "busFactor": 2,
    "keyOwners": [...]
  },
  "warnings": [
    "Low bus factor: project depends on 2 or fewer contributors"
  ]
}

AI can use this context to:

• Adjust behavior based on file risk levels
• Understand ownership before suggesting changes
• Detect hidden dependencies
• Provide context-aware suggestions

Development

# Install dependencies
npm install

# Build
npm run build

# Run locally
npm run dev reflect
npm run dev context --json

# Link for global testing
npm link

Roadmap

• Phase 1: Personal retrospectives (reflect, churn, heatmap)
• Phase 2: Contributor onboarding (blame-map, scars, couples)
• Phase 3: Project management (report, blockers, evolution)
• AI Context (context command, SKILL.md auto-generation)
• File filtering (built-in ignores, .gitspectrc config)
• Phase 4: Enhanced AI integration (agent prompts, .cursorrules generation)
• LLM integration (optional local/cloud models)

License

MIT © tscburak