Robotic Capture-the-Flag on CTF-2
5Success CountEnvironment-Grounded Multi-Agent Workflow
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Environment-Grounded Multi-Agent WorkflowModel=llama-3.3-70b-instruct, Approach=Proposed2026.03 | 5 | 0 | |
| Environment-Grounded Multi-Agent WorkflowModel=deepseek-v3.2, Approach=Proposed2026.03 | 1 | 4 | |
| Environment-Grounded Multi-Agent WorkflowModel=gemma-3-27b-it, Approach=Proposed2026.03 | 0 | 5 | |
| Environment-Grounded Multi-Agent WorkflowModel=hermes-2-pro-llama-3-8b, Approach=Proposed2026.03 | 0 | 5 | |
| HackSynthModel=llama-3.3-70b-instruct, Approach=HackSynth [20]2026.03 | 0 | 5 | |
| HackSynthModel=deepseek-v3.2, Approach=HackSynth [20]2026.03 | 0 | 5 | |
| HackSynthModel=gemma-3-27b-it, Approach=HackSynth [20]2026.03 | 0 | 5 | |
| HackSynthModel=hermes-2-pro-llama-3-8b, Approach=HackSynth [20]2026.03 | 0 | 5 |