How icebergn2026/artex-ld works
artex-ld is an AI-assisted penetration-testing platform. An operator configures a task and target in the web console; the backend manages the task, then its planner creates work for workers to execute, with findings and evidence available for review and reporting. The README also documents configurable scenarios, intent approval, custom executors, and finding-triggered follow-up tasks or retests. The graph combines these documented capabilities with sampled engine and preference code; several subsystem links are documented rather than verified in source excerpts.
Main entry points: worker.go Worker · engine.go Execution Engine · findings.go Findings · page.tsx Web Console
Components 21
Operator Console
- Web Console
web/src/app/(main)/function/tasks/page.tsx - Web API Client
web/src/lib/api/tasks.ts - UI Preferences
web/src/lib/preferences/preferences-storage.ts
Task Control
- HTTP API
server/server.go - Task Manager
server/manager.go - Execution Engine
server/engine.go
Agent Execution
- Planner
agent/planner.go - Worker
agent/worker.go - Agent Tools
agent/tools.go
Review and Findings
- Tool Guard
guard/guard.go - Intent Approval
intercept/intercept.go - Findings
db/findings.go - Finding Orchestration
server/finding_workflow.go - Finding Reports
report/findings.go
State and Integrations
- Task Registry
db/tasks.go - Exploration Graph
db/exploration.go - Traffic Evidence
traffic/traffic.go - Evidence Store
evidence/store.go
Other components
- Operator
- LLM Provider
- MCP Services
Connections 22
- Operator → Web Console: configures tasks
- Web Console → Web API Client: requests task data
- Web API Client → HTTP API: calls API
- HTTP API → Task Manager: handles requests
- Task Manager → Task Registry: manages tasks
- Execution Engine → Task Registry: dispatches work
server/engine.go - Execution Engine → Planner: runs planner
- Execution Engine → Worker: dispatches intents
server/engine.go - Planner → LLM Provider: requests plans
- Worker → LLM Provider: executes agent turns
- Worker → Agent Tools: uses tools
- Worker → Tool Guard: gates tool calls
- Tool Guard → Intent Approval: checks review policy
- Worker → Exploration Graph: records discoveries
- Worker → Findings: reports findings
- Findings → Finding Orchestration: triggers follow-up
- Finding Orchestration → Execution Engine: creates follow-up work
- Findings → Finding Reports: feeds reports
- Worker → Traffic Evidence: captures traffic
- Worker → Evidence Store: stores evidence
- Worker → MCP Services: uses integrations
- Operator → UI Preferences: sets preferences
web/src/lib/preferences/preferences-storage.ts