How paulinotech1/strix works
Strix is an autonomous AI penetration-testing product, exposed through a local CLI and interactive interfaces, with a separate cloud-facing workflow. A scan takes targets and configuration, prepares a runtime, orchestrates agents and tools to investigate and validate vulnerabilities, then records and presents findings. The README also describes multi-agent collaboration, exploit validation, autofix, and compliance reporting. The graph preserves those capabilities; where implementation was not sampled, relationships are omitted or marked unverified rather than inferred from names alone.
Main entry points: execution.py Agent execution · state.py Run and finding state · main.py CLI entry · factory.py Agent construction
Components 23
User interfaces
- CLI entry
strix/interface/main.py - Interactive TUI
strix/interface/tui/backend/controller.py - Run viewer
strix/interface/viewer/server.py
Scan orchestration
- Configuration
strix/config/loader.py - Targets and inputs
strix/core/inputs.py - Scan runner
strix/core/runner.py - Agent execution
strix/core/execution.py - Container runtime
strix/runtime/session_manager.py - Model providers
strix/config/models.py
Agent capabilities
- Agent construction
strix/agents/factory.py - Agent coordination
strix/core/agents.py - Pentest tools
strix/tools/proxy/tools.py - MCP connections
strix/tools/mcp/client.py - Agent skills
strix/skills/__init__.py
Findings and reports
- Run and finding state
strix/report/state.py - Finding deduplication
strix/report/dedupe.py - Report generation
strix/report/writer.py - Scan telemetry
strix/telemetry/posthog.py
Cloud and integrations
- Cloud scan interface
strix/interface/cloud/runner.py - Cloud HTTP API
strix/interface/cloud/http.py
Other components
- User
- LLM provider
- Target application
Connections 24
- User → CLI entry: starts scan
README.md - CLI entry → Configuration: loads settings
strix/interface/main.py - CLI entry → Targets and inputs: prepares targets
strix/interface/main.py - CLI entry → Scan runner: launches scan
- Scan runner → Agent execution: runs agents
- Agent execution → Targets and inputs: uses targets
strix/core/execution.py - Agent execution → Agent coordination: coordinates agents
strix/core/execution.py - Agent execution → Model providers: configures model
- Agent execution → Container runtime: uses runtime
- Agent construction → Pentest tools: registers tools
strix/agents/factory.py - Agent construction → MCP connections: enables MCP
strix/agents/factory.py - Agent construction → Agent skills: loads skills
strix/agents/factory.py - User → Target application: assesses
README.md - Pentest tools → Target application: tests target
- MCP connections → LLM provider: calls tools
- Agent execution → Run and finding state: records results
- Run and finding state → Finding deduplication: deduplicates findings
strix/report/dedupe.py - Run and finding state → Report generation: writes reports
strix/report/state.py - Run and finding state → Scan telemetry: sends scan data
strix/telemetry/posthog.py - Run and finding state → Run viewer: serves run data
strix/interface/viewer/server.py - User → Interactive TUI: steers scan
- Interactive TUI → Run viewer: opens viewer
strix/interface/tui/backend/controller.py - User → Cloud scan interface: requests cloud scan
README.md - Cloud scan interface → Cloud HTTP API: uses API