FreeLLMN1nEmAn/sapper-v2: Autonomous pentest legion — AI commander + worker swarm + layered fact-intent graph. 3rd Te
The competition featured 634 teams and ran for eight hours unattended, during which sapper‑v2 achieved an 80.18/100 score and captured 85 flags, 25 seconds fast
The competition featured 634 teams and ran for eight hours unattended, during which sapper‑v2 achieved an 80.18/100 score and captured 85 flags, 25 seconds faster than the runner‑up.
The system uses a multi‑agent orchestration framework that can swap any AI model for each role, such as Claude Sonnet, GLM‑5.3, DeepSeek V4.1 Flash, or Code‑Buddy.
During the test the legion consumed 4.5 billion tokens while staying within a 1 billion token limit, demonstrating efficient token management at scale.
Developers describe sapper‑v2 as a “legion of AIs” that coordinates a commander, generals, soldiers and a watchdog, storing knowledge in a layered fact‑intent graph.
Ce que cela signifie pour vous
The result shows that coordinated multi‑agent systems can outperform single large models in complex security tasks, offering a template for future autonomous testing tools. Users seeking free or locally run AI can adapt the framework with open models like GLM‑5.3, while paid‑model users may see value in the orchestration approach for scaling security operations.
Sources
- N1nEmAn/sapper-v2: Autonomous pentest legion — AI commander + worker swarm + layered fact-intent graph. 3rd Tencent Intelligent Attack-Defense online champion (634 teams · 8h unattended · 1B tokens be, GitHub trending, 30 Oct 2026
Rédigé automatiquement à partir des sources ci‑dessus, puis vérifié par notre code : chaque nombre et chaque nom doivent y figurer, et aucune phrase ne doit copier une source. Une erreur ? Signalez‑la.
Quelle IA pouvez‑vous utiliser gratuitement aujourd’hui ?
Tous les modèles IA gratuits, classés par votes humains anonymes, avec la mémoire et la vitesse dont ils ont besoin sur votre propre GPU.