PyRIT Command-Line Frontends

PyRIT Command-Line Frontends#

PyRIT provides two command-line interfaces for running AI red teaming scenarios:

pyrit_scan - Single-Command Execution#

pyrit_scan is designed for automated, non-interactive scenario execution. It’s ideal for:

  • CI/CD pipelines and automated testing workflows

  • Batch processing multiple scenarios with scripts

  • One-time security assessments

  • Reproducible test runs with exact parameters

Each invocation runs a single scenario with specified parameters and exits, making it perfect for automation where you need clean, scriptable execution with predictable exit codes.

Key characteristics:

  • Loads PyRIT modules fresh for each execution

  • Runs one scenario per command

  • Exits with status code (0 for success, non-zero for errors)

  • Output can be easily captured and parsed

Documentation: 1_pyrit_scan.ipynb

pyrit_shell - Interactive Session#

pyrit_shell is an interactive REPL (Read-Eval-Print Loop) for exploratory testing. It’s ideal for:

  • Interactive scenario development and debugging

  • Rapid iteration and experimentation

  • Exploring multiple scenarios without reload overhead

  • Session-based result tracking and comparison

The shell loads PyRIT modules once at startup and maintains a persistent session, allowing you to run multiple scenarios quickly and review their results interactively.

Key characteristics:

  • Fast subsequent executions (modules loaded once)

  • Session history of all runs

  • Interactive result exploration and printing

  • Persistent context across multiple scenario runs

  • Tab completion and command help

Documentation: 2_pyrit_shell.md

Contents