What problem does it solve? Manual penetration testing requires coordinating dozens of security tools, interpreting their output, and writing reports, which is slow and error-prone. Guardian orchestrates 19+ security tools with AI providers (GPT-4, Claude, Gemini, OpenRouter) to plan, execute, and document security assessments automatically. ## Core Features & Use Cases - AI-Driven Workflows: Run predefined workflows (web_pentest, network, recon, autonomous) where an LLM selects tools, adapts to findings, and reasons about next steps. - Custom Workflows and Tools: Define YAML-based workflows with per-step AI instructions and register custom Python tool classes for specialized scanners. - Report Generation: Produce markdown, HTML, or JSON pentest reports from session data with evidence capture and severity filtering. - Use Case: A security engineer runs python -m cli.main workflow run --name web_pentest --target example.com --provider claude to execute HTTP discovery, vulnerability scanning, SQL injection and XSS testing, then generates an HTML report of all findings. ## Quick Start Set your AI provider API key as an environment variable, then ask the agent to run the Guardian web_pentest workflow against your authorized target and generate a markdown report of the findings.
