cytoscnpy 1.2.1

High-performance dead code elimination analysis tool for Python.

CytoScnPy - High-Performance Python Static Analysis

A fast static analysis tool for Python codebases, powered by Rust with hybrid Python integration. Detects dead code, security vulnerabilities (including taint analysis), and code quality issues with extreme speed. Code quality metrics are also provided.

Why CytoScnPy?

• Blazing Fast: Faster in dead code detection.
• Memory Efficient: Uses less memory.
• Comprehensive: Dead code, secrets, security, taint analysis, quality metrics
• Framework Aware: Flask, Django, FastAPI, Pydantic, Azure Functions
• Benchmarked: Continuous benchmarking with 135-item ground truth suite

Installation

Linux / macOS:

# Install
curl -fsSL https://raw.githubusercontent.com/djinn09/CytoScnPy/main/install.sh | bash

Windows (PowerShell):

# Install
irm https://raw.githubusercontent.com/djinn09/CytoScnPy/main/install.ps1 | iex

Via Pip:

pip install cytoscnpy

From Source:

git clone https://github.com/djinn09/CytoScnPy.git
cd CytoScnPy
pip install maturin
maturin develop -m cytoscnpy/Cargo.toml

MCP Server (for AI Assistants)

CytoScnPy includes an MCP server for AI assistant integration:

# Start MCP server (after pip install)
cytoscnpy mcp-server

For Claude Desktop, Cursor, or GitHub Copilot configuration, see the MCP Server Documentation.

Features

• Dead Code Detection: Unused functions, classes, imports, and variables with cross-module tracking.
  • Cascading Detection: Methods inside unused classes are automatically flagged as unused.
  • Auto-Fix: Remove dead code automatically with --fix (preview by default, use --apply to execute).
• Clone Detection: Find duplicate code with --clones.
• Security Analysis: Taint analysis (SQLi, XSS), secret scanning (API keys, suspicious variables), and dangerous code patterns (eval, exec).
• Code Quality Metrics: Cyclomatic complexity, Halstead metrics, Maintainability Index, and raw metrics (LOC, SLOC).
• Framework Support: Native understanding of Flask, Django, FastAPI, Pydantic, and Azure Functions v2 patterns.
• Smart Heuristics: Handles dataclasses, __all__ exports, visitor patterns, and dynamic attributes intelligently.
• Cross-File Detection: Tracks symbol usage across the entire codebase, including nested packages and complex import chains, to ensure code used in other modules is never incorrectly flagged.

Usage

Command Line

cytoscnpy [PATHS]... [OPTIONS]

Examples:

# Dead code analysis
cytoscnpy .                                     # Analyze current directory
cytoscnpy /path/to/project --json               # JSON output for CI/CD

# Security checks (short flags: -s, -d, -q)
cytoscnpy . --secrets --danger --quality
cytoscnpy . -s -d -q                        # Same with short flags

# Confidence threshold (0-100)
cytoscnpy . --confidence 80

# Path filtering
cytoscnpy . --exclude-folder venv --exclude-folder build
cytoscnpy . --include-folder specific_venv      # Override defaults
cytoscnpy . --include-tests

# Jupyter notebooks
cytoscnpy . --include-ipynb --ipynb-cells

# Clone detection (find duplicate code)
cytoscnpy . --clones --clone-similarity 0.8

# Auto-fix dead code (preview first, then apply)
cytoscnpy . --fix                    # Preview changes (dry-run by default)
cytoscnpy . --fix --apply            # Apply changes
cytoscnpy . --fix -a                 # Apply changes (short flag)

# Generate HTML report (quality auto-enabled; add --secrets --danger for security)
cytoscnpy . --html --secrets --danger

Options:

Flag	Description
-c, --confidence <N>	Set confidence threshold (0-100)
-s, --secrets	Scan for API keys, tokens, credentials
-d, --danger	Scan for dangerous code + taint analysis
-q, --quality	Scan for code quality issues
-n, --no-dead	Skip dead code detection (security/quality only)
--html	Generate HTML report (auto-enables quality)
--json	Output results as JSON
-v, --verbose	Enable verbose output for debugging
--quiet	Quiet mode: summary only, no tables
--include-tests	Include test files in analysis
--exclude-folder <DIR>	Exclude specific folders
--include-folder <DIR>	Force include folders
--include-ipynb	Include Jupyter notebooks
--ipynb-cells	Report findings per notebook cell
--clones	Detect duplicate code
--clone-similarity <N>	Clone similarity threshold (0.0-1.0)
--fix	Preview dead code removal (dry-run by default)
-a, --apply	Apply --fix changes to files

CI/CD Gate Options:

Flag	Description
--fail-threshold <N>	Exit code 1 if unused code % > N
--max-complexity <N>	Exit code 1 if any function complexity > N
--min-mi <N>	Exit code 1 if maintainability index < N
--fail-on-quality	Exit code 1 if any quality issues found
--max-nesting <N>	Exit code 1 if any block nesting > N
--max-args <N>	Exit code 1 if any function has > N args
--max-lines <N>	Exit code 1 if any function has > N lines

Full CLI Reference: See docs/CLI.md for complete command documentation.

Metric Subcommands

cytoscnpy raw .                    # Raw Metrics (LOC, SLOC, Comments)
cytoscnpy cc .                     # Cyclomatic Complexity
cytoscnpy hal .                    # Halstead Metrics
cytoscnpy mi .                     # Maintainability Index
cytoscnpy stats . --all            # Full project report (secrets, danger, quality)
cytoscnpy stats . --all -o report.md  # Save report to file
cytoscnpy files .                  # Per-file metrics table

Tip: Add --json for machine-readable output, --exclude-folder <DIR> to skip directories globally, or --ignore <PATTERN> for subcommand-specific glob filtering.

Feature Flags

The crate supports experimental features that can be enabled at compile time:

Feature	Description
cfg	Enables Control Flow Graph (CFG) construction and behavioral validation for clone detection

To build with a feature enabled:

cargo build --features cfg

⚙️ Configuration

Create .cytoscnpy.toml (uses [cytoscnpy]) or add to pyproject.toml (uses [tool.cytoscnpy]):

.cytoscnpy.toml example:

[cytoscnpy]
# General Settings
confidence = 60  # Minimum confidence threshold (0-100)
exclude_folders = ["venv", ".tox", "build", "node_modules", ".git"]
include_folders = ["src", "tests"]  # Optional: whitelist folders
include_tests = false  # Note: include_ipynb and ipynb_cells are CLI-only (use flags)

# Analysis Features
secrets = true
danger = true
quality = true

# Fail Threshold (exit code 1 if exceeded)
fail_threshold = 10.0  # Fail if >10% of code is unused
# fail_threshold = 0.0  # Zero tolerance: fail on any unused code

# Code Quality Thresholds
max_lines = 100       # Max lines per function
max_args = 5          # Max arguments per function
complexity = 10       # Max cyclomatic complexity
nesting = 4           # Max indentation depth
min_mi = 65.0         # Minimum Maintainability Index
ignore = ["R001"]     # Ignore specific rule IDs

# Advanced Secret Scanning
[cytoscnpy.secrets_config]
entropy_enabled = true
entropy_threshold = 4.5  # Higher = more random (API keys usually >4.0)
min_length = 16          # Min length to check for entropy
scan_comments = true     # Scan comments for secrets
skip_docstrings = false  # Skip docstrings in entropy scanning
min_score = 50           # Minimum confidence score (0-100)
suspicious_names = ["db_config", "oauth_token"] # Add custom suspicious variable names

# Custom Secret Patterns
[[cytoscnpy.secrets_config.patterns]]
name = "Slack Token"
regex = "xox[baprs]-([0-9a-zA-Z]{10,48})"
severity = "HIGH"

Note: Notebook options (include_ipynb, ipynb_cells) are currently CLI-only but will be added to the configuration file in a future release.

CI/CD Quality Gates

Configure quality gates for CI/CD pipelines. Set thresholds and the CLI exits with code 1 if exceeded.

CLI Flags:

# Unused code percentage gate
cytoscnpy . --fail-threshold 5  # Fail if >5% unused

# Complexity gate
cytoscnpy . --max-complexity 10  # Fail if any function >10

# Maintainability Index gate
cytoscnpy . --min-mi 40  # Fail if MI <40

# Quiet mode for clean CI output
cytoscnpy . --fail-threshold 5 --quiet

Priority: CLI flag > config file > environment variable > default

Environment Variable: CYTOSCNPY_FAIL_THRESHOLD=5.0

Performance

Accuracy (Benchmark Suite: 135 items)

Detection Type	Precision	Recall	F1 Score
Classes	0.73	0.79	0.76
Functions	0.71	0.74	0.73
Methods	0.86	0.93	0.89
Imports	0.67	0.40	0.50
Variables	0.30	0.15	0.20
Overall	0.71	0.64	0.68

See benchmark/README.md for detailed comparison against Vulture, Flake8, Pylint, Ruff, and others.

Architecture

See cytoscnpy/README.md for detailed architecture and technology stack information.

Testing

See CONTRIBUTING.md for testing instructions.

Contributing

See CONTRIBUTING.md for development setup and guidelines.

License

Apache-2.0 License - see License file for details.

Links

• Documentation: CytoScnPy
• PyPI: PyPi
• VS Code Extension: VS Code Marketplace
• Rust Core Documentation: cytoscnpy/README.md
• Benchmarks & Accuracy: benchmark/README.md
• Roadmap: ROADMAP.md
• Contributing: CONTRIBUTING.md

References

CytoScnPy's design and implementation are inspired by:

• Skylos
• Radon