Black Box Unlock Mischief. Mayhem. Merge conflicts. Exposed. Code forensics tool based on Adam Tornhill's "Your Code as a Crime Scene". Key insight: 2-8% of files cause 60-90% of defects. Built for AI coding agents: forensic signals as MCP tools and a Claude Code plugin, so reviews and refactors are prioritized by evidence.

For agents (MCP + plugin)

uv tool install black-box-unlock   # provides bbu and bbu-mcp

Register the MCP server in Claude Code (.mcp.json):

{ "mcpServers": { "black-box-unlock": { "command": "bbu-mcp" } } }

Tools: get_hotspots, get_file_forensics, get_coupled_files, get_ownership, get_ci_failures, get_flaky_steps, xray_file.

The Claude Code plugin in this repo adds /analyze, /hotspots, a git-forensics agent, and an ambient coupling guard that warns when you edit one half of a temporally coupled file pair. Install it via the self-hosted marketplace:

/plugin marketplace add michael-denyer/black-box-unlock
/plugin install black-box-unlock@black-box-unlock

Both bbu and bbu-mcp must be on PATH for the plugin and MCP server to work.

CLI

Installation

uv pip install -e .

CI failure analysis additionally uses the gh CLI when available (skip with --no-ci).

Usage

# Analyze last 30 days of git history, output JSON
bbu analyze-repo --days=30

# Generate interactive HTML report
bbu analyze-repo --days=30 --output=html > report.html

# Adjust coupling detection threshold (default 0.3)
bbu analyze-repo --min-coupling=0.5 --output=html > report.html

# Skip CI failure analysis (faster, no GitHub access needed)
bbu analyze-repo --no-ci --output=html > report.html

# Analyze a different repository
bbu analyze-repo --repo /path/to/repo --output=html > report.html

# Per-function churn for one file (Tornhill's X-Ray)
bbu xray src/hot_file.py --days 365

Features

Does the ranking actually predict bugs?

Measured with bbu validate (split-history: rank hotspots on the older half, count bug-fix commits in the newer half): median Spearman rho 0.46 across six real repos (click, flask, pydantic, rich, fastapi, httpx); the top 10% of ranked files attracted a median 46% of subsequent bug-fix touches — uniform would be 10%. Method, per-repo numbers, and limitations: docs/VALIDATION.md.

bbu validate --repo /path/to/repo --days 730

HTML report

The HTML report includes three interactive views:

• Table - Sortable file metrics with severity coloring
• Hotspots - Plotly treemap showing file churn by directory
• Coupling - Cytoscape.js network graph of temporal coupling

The HTML report is feature-frozen; new signals land in JSON and MCP only.

flowchart LR
    Git[Git History] --> Analyze[bbu analyze-repo]
    CI[GitHub Actions] --> Analyze
    Analyze --> JSON[JSON Output]
    Analyze --> HTML[HTML Report]
    HTML --> Treemap[Hotspot Treemap]
    HTML --> Graph[Coupling Graph]

Architecture

See docs/ARCHITECTURE.md for full details.

src/black_box_unlock/
├── cli.py              # Typer CLI
├── complexity.py       # Indentation-depth complexity proxy
├── analysis.py         # Orchestration
├── core/               # Pydantic models, exceptions, logging
├── git/                # Churn, coupling, ownership, defects, log extraction
├── cicd/               # CI/CD forensics (build failures, flaky steps via gh CLI)
└── visualization/      # HTML, treemap, coupling graph (frozen)

Development

# Run tests
uv run pytest -v

# Lint and format
uv run ruff check . && uv run ruff format .

# Verbose output for debugging
bbu --verbose analyze-repo

License

MIT