Tool-first Execution Loop
The model actively calls terminal commands for evidence and iterates until a reproducible result is reached.
Replace vague chat outputs with actionable evidence. Route model reasoning to real Kali tools and keep "analyze -> execute -> verify -> conclude" in one continuous session.
Input known context and task constraints to build the initial analysis frame.
Use low-level Kali tools and collect real command evidence.
Adjust strategy from output feedback and continue iteration.
Produce verifiable conclusions and next actions to close the loop.
No fluff, just outcomes. Collect terminal-true evidence via low-level execution with fully traceable steps.
The model actively calls terminal commands for evidence and iterates until a reproducible result is reached.
Automatically handles upstream unavailability or HTTP 502, then switches to available model nodes to keep analysis continuous.
Data stays scoped with per-user isolation.
Realtime heartbeat checks quickly remove offline inference nodes.
Cross-domain collaboration to build practical security tooling.
Practical contest performance, key events, and reproducible writeups.
Contest Hall
Review recent online contests, weekly matches, and training events in one place.
Open Contest HallUNICTF
Showcase UNICTF performance, key breakthroughs, and post-match review notes.
View UNICTF PageAwards
Summarize rankings, awards, and milestone achievements across competitions.
View Awards PageFrom low-level design to practical methodology, understand how this platform is built.
Paper 01
Explain system architecture, session boundaries, execution chain, and security baseline.
Read PaperPaper 02
Analyze failover and round-robin routing, plus reliability and fault-tolerance strategy.
Read PaperPaper 03
Describe the step-by-step workflow from input to conclusion with automation constraints.
Read PaperBusiness cooperation, feedback, and vulnerability discussion are welcome.