Headless Mode

loop --print "your prompt here"

# Explain a file
loop --print "explain src/main.rs"

# Generate code
loop --print "write a Python script that converts CSV to JSON" > convert.py

# Analyze a project
loop --print "list all TODO comments in this project"

# provider/model in one flag
loop --model soket/qwen3-30b --print "review this code for bugs"

# Or provider and model separately (both are required together)
loop --provider openai --model gpt-4o --print "summarize the README"

# Custom system prompt
loop --system-prompt "You are a security auditor" --print "audit src/auth.rs"

# Append to system prompt
loop --append-system-prompt "Always respond in JSON" --print "list all functions in src/lib.rs"

# Skip context files
loop --no-context-files --print "what is 2+2"

# Set working directory
loop --cwd /path/to/project --print "explain this project"

# Save to file
loop --print "generate a .gitignore for a Rust project" > .gitignore

# Pipe to another command
loop --print "list all API endpoints in this project as JSON" | jq '.endpoints[]'

#!/bin/bash
for file in src/*.rs; do
  echo "=== Reviewing $file ==="
  loop --print "review $file for potential bugs and suggest fixes"
  echo
done

# .github/workflows/review.yml
- name: AI Code Review
  run: |
    loop --print "review the changes in this PR for bugs and security issues" \
      > review-output.md

loop --print "explain this crate" \
  --trace-name bench \
  --trace-tag swe-bench \
  --trace-metadata task=django__django-1234 \
  --trace-env benchmark