COMMAND: python3 -I review.py --version
2.1.1
EXIT_CODE=0
COMMAND: python3 -I review.py --help
usage: review.py [-h] [--version]
                 {preflight,bootstrap,doctor,scan,ai,auth-check,demo,compare} ...

Security Review Project: install scanners, scan a Git commit, and optionally
review with Claude Code.

positional arguments:
  {preflight,bootstrap,doctor,scan,ai,auth-check,demo,compare}
    preflight           Check fresh-host Python/Git/venv prerequisites without
                        network access
    bootstrap           Install pinned scanners locally without sudo
    doctor              Check Python/Git and installed scanner versions;
                        Claude is optional
    scan                Run real scanners on a clean, committed repository
                        snapshot
    ai                  Run bounded discovery and independent verification on
                        an existing scan
    auth-check          Check local Claude CLI capabilities/auth; no source
                        upload or model request
    demo                Run vulnerable/fixed fixtures and real-scanner
                        acceptance; never install vulnerable dependencies
    compare             Compare findings between two scan reports

options:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
EXIT_CODE=0
COMMAND: python3 -I review.py preflight
{
  "status": "HOST_PREREQUISITES_PASSED",
  "python": "3.13.5",
  "platform": "linux-x86_64",
  "git": "git version 2.47.3",
  "venv_with_pip": true,
  "note": "No network, scanner, database, Claude login or model request has been verified."
}
EXIT_CODE=0
COMMAND: python3 -I review.py auth-check --help
usage: review.py auth-check [-h] --auth {subscription,api}

options:
  -h, --help            show this help message and exit
  --auth {subscription,api}
EXIT_CODE=0
COMMAND: python3 -I review.py ai --help
usage: review.py ai [-h] --run RUN --auth {subscription,api}
                    [--allow-code-upload] [--model MODEL]
                    [--budget-usd BUDGET_USD] [--max-turns MAX_TURNS]
                    [--ai-timeout AI_TIMEOUT]

options:
  -h, --help            show this help message and exit
  --run RUN
  --auth {subscription,api}
                        Explicit credential/billing mode; never automatically
                        falls back
  --allow-code-upload   Explicitly consent to sending selected source excerpts
                        to Anthropic
  --model MODEL         Claude model alias or exact model ID; set an exact ID
                        for evaluated deployments
  --budget-usd BUDGET_USD
                        API-only total CLI budget (default 4 USD), split
                        across two calls; not a subscription quota
  --max-turns MAX_TURNS
                        Turn limit per Claude call (1-20; default 3)
  --ai-timeout AI_TIMEOUT
                        Wall-clock seconds per Claude call (1-3600; default
                        240)
EXIT_CODE=0
