CLI Reference

CLI Reference

GraphOptim provides a full-featured command-line interface.

Global Options

graphoptim --version    Show version
graphoptim --help       Show help

---

graphoptim analyze

Read-only code inspection.

graphoptim analyze PATH [OPTIONS]

Arguments:

Argument	Description
PATH	Python file or directory to analyze

Options:

Option	Default	Description
--verbose, -v	false	Show detailed output including node/edge counts
--output, -o	—	Save report to file
--format	text	Output format: text or json
--threshold	0.7	Betweenness centrality threshold

Examples:

# Analyze a single file
graphoptim analyze myfile.py

# Verbose analysis with all metrics
graphoptim analyze myfile.py --verbose

# JSON output
graphoptim analyze myfile.py --format json

# Save report to file
graphoptim analyze myfile.py -o report.txt

# Analyze entire project
graphoptim analyze ./src --verbose

---

graphoptim optimize

Apply graph-theoretic optimization passes.

graphoptim optimize PATH [OPTIONS]

Arguments:

Argument	Description
PATH	Python file or directory to optimize

Options:

Option	Default	Description
--dry-run	true	Preview only (default)
--inplace	false	Modify file in-place (creates .bak backup)
--output, -o	—	Write optimized code to file/directory
--passes, -p	auto	Comma-separated list of passes
--budget	0.6	Risk budget for knapsack selection (0.0-1.0)

Available Passes:

Pass Name	Description
dead_code	Remove unreachable code
path_shortener	Merge duplicate branches
centrality	Add advisory comments at bottlenecks

Examples:

# Preview optimization
graphoptim optimize myfile.py

# Apply specific passes
graphoptim optimize myfile.py -p dead_code,path_shortener --inplace

# Optimize with low-risk budget
graphoptim optimize myfile.py --budget 0.3 --inplace

# Optimize to new directory
graphoptim optimize ./src -o ./src_optimized

---

graphoptim benchmark

Run the empirical benchmark study.

graphoptim benchmark [OPTIONS]

Options:

Option	Default	Description
--samples, -n	100	Number of benchmark problems
--models, -m	claude,gpt4o,gemini	Comma-separated model list
--dataset	humaneval	Dataset: humaneval or mbpp
--output-dir, -o	benchmark_results	Output directory

Required Environment Variables:

export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AI..."

Examples:

# Quick benchmark with Claude only
graphoptim benchmark --samples 20 --models claude

# Full benchmark
graphoptim benchmark --samples 100 --models claude,gpt4o,gemini

# Using MBPP dataset
graphoptim benchmark --dataset mbpp --samples 50

---

graphoptim config

Show or set configuration.

graphoptim config ACTION [KEY] [VALUE]

Actions:

Action	Description
show	Display current configuration
set	Set a configuration value

Examples:

# Show all settings and API key status
graphoptim config show

# Set an API key (suggests using env vars)
graphoptim config set anthropic_api_key sk-ant-...