AI Analysis

PCU integrates with AI CLI tools to provide intelligent dependency analysis, security assessments, and update recommendations.

Overview

AI analysis enhances PCU's capabilities by providing:

Impact Analysis: Understand how updates affect your codebase
Security Assessment: Get AI-powered security vulnerability analysis
Compatibility Checking: Detect potential breaking changes
Update Recommendations: Receive intelligent suggestions for safe updates

Supported AI Providers

PCU automatically detects and uses available AI CLI tools in the following priority order:

Provider	Priority	Capabilities
Gemini	100	Impact, Security, Compatibility, Recommendations
Claude	80	Impact, Security, Compatibility, Recommendations
Codex	60	Impact, Compatibility, Recommendations
Cursor	40	Impact, Recommendations

If no AI providers are available, PCU automatically falls back to a rule-based analysis engine that provides basic dependency analysis using predefined rules.

Commands

Check Available AI Providers

View which AI tools are available on your system:

pcu ai

This command displays:

Available AI CLI tools detected on your system
Version information for each provider
The best available provider that will be used for analysis

AI Command Options

Name
--status
Type
boolean
Description
Show status of all AI providers (default behavior)
Name
--test
Type
boolean
Description
Test AI analysis with a sample request to verify provider connectivity
Name
--cache-stats
Type
boolean
Description
Show AI analysis cache statistics including hit rate and size
Name
--clear-cache
Type
boolean
Description
Clear the AI analysis cache to free up space or reset cached responses

pcu ai
pcu ai --status

AI-Powered Update

Update dependencies with AI-powered analysis:

pcu update --ai
pcu u --ai

The AI-enhanced update provides:

Intelligent risk assessment for each update
Breaking change detection with explanations
Security vulnerability identification
Recommended update order

AI-Powered Analysis

Analyze a specific package update with AI assistance:

pcu analyze react
pcu a lodash

The analyze command uses the default catalog by default. You can specify a different catalog as the first argument: pcu analyze my-catalog react

Analysis Types

Impact Analysis

Evaluates how a dependency update will affect your project:

Identifies all workspace packages using the dependency
Analyzes API changes between versions
Estimates migration effort required
Suggests testing focus areas

Security Analysis

Provides security-focused assessment:

Identifies known vulnerabilities in current version
Checks for security fixes in new version
Evaluates security-sensitive package updates
Recommends security best practices

Compatibility Analysis

Checks for potential compatibility issues:

Detects breaking API changes
Identifies peer dependency conflicts
Checks Node.js version compatibility
Validates TypeScript compatibility

Recommendations

Generates actionable recommendations:

Suggests optimal update order
Recommends version ranges
Identifies packages to update together
Provides rollback strategies

Fallback Behavior

When AI providers are unavailable, PCU uses a built-in rule-based analysis engine:

Rule-Based Analysis Features

Version Jump Assessment: Evaluates risk based on semver changes
Known Breaking Patterns: Detects breaking changes for popular packages (React, TypeScript, ESLint, etc.)
Security-Sensitive Packages: Flags security-related packages for careful review
Effort Estimation: Provides migration effort estimates

Risk Levels

Level	Description
Low	Patch updates, typically safe to apply
Medium	Minor updates or large minor version jumps
High	Major version updates with breaking changes
Critical	Multiple major version jumps or pre-release versions

Configuration

Environment Variables

Name
GEMINI_PATH
Type
string
Description
Custom path to Gemini CLI executable
Name
CLAUDE_PATH
Type
string
Description
Custom path to Claude CLI executable
Name
CODEX_PATH
Type
string
Description
Custom path to Codex CLI executable
Name
CURSOR_PATH
Type
string
Description
Custom path to Cursor CLI executable

Detection Methods

PCU uses multiple strategies to detect AI providers:

Environment Variables: Check custom path variables
PATH Lookup: Use which command to find executables
Known Paths: Check common installation locations
Application Paths: Check for GUI applications (e.g., Cursor.app)

Usage Examples

Safe Update Workflow

# Check for updates with AI analysis
pcu check

# Analyze specific high-risk updates
pcu analyze react 19.0.0
pcu analyze typescript 5.5.0

# Perform AI-guided interactive update
pcu update --ai --interactive --create-backup

CI/CD Integration

# Run AI analysis in CI with JSON output
pcu update --ai --dry-run --format json > ai-analysis.json

# Check for high-risk updates
pcu analyze default typescript --format json | jq '.riskLevel'

Batch Analysis

# Update all packages with AI recommendations
pcu update --ai --target minor

# Review major updates with AI analysis
pcu update --ai --target latest --confirm-major

Best Practices

When to Use AI Analysis

Major Version Updates: Always use AI analysis for major version bumps
Security-Sensitive Packages: Use for auth, crypto, and session packages
Large Codebases: AI helps identify affected areas across monorepos
Breaking Change Detection: AI provides detailed breaking change explanations

Performance Considerations

AI analysis adds processing time compared to standard updates
Use --dry-run to preview AI recommendations without applying changes
Consider using rule-based fallback for faster CI/CD pipelines when AI is not critical

Combining with Other Features

# AI analysis with security scanning
pcu update --ai && pcu security

# AI analysis with backup
pcu update --ai --create-backup

# AI analysis with specific target
pcu update --ai --target minor --exclude "*-dev"