byerlikaya-red-team

SKILLWorkflowCommunity
v0.0.0byerlikayaMITAktualisiert vor 9 TQuelle →

Attacker's-eye test of LLM/agent defenses: instruction hijacking, data exfiltration and tool abuse through untrusted content; verifies whether the defense actually holds.

Community-submitted skill. Not yet reviewed by the Forge team. Full prompt content may not be available.Request review →
22Repo-Sterne
1Clients
1Formate
vor 9 TLetzte Aktualisierung
Skill
Autorbyerlikaya
Version0.0.0
LizenzMIT
KategorieWorkflow
Formateskill.md
PromptNicht veröffentlicht
Kompatibilität
Claude✓ Unterstützt
Cursor
Copilot
ChatGPT
Gemini
Über

Attacker's-eye test of LLM/agent defenses: instruction hijacking, data exfiltration and tool abuse through untrusted content; verifies whether the defense actually holds.

Schlagwörter
skillclaude