nuke-blitz

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md, Nuke Blitz consumes the caller-provided task (inline text or a file path) to construct the contract and to feed the “task text” into the fresh validator, meaning outsider-authored free text can directly enter LLM prompts at CALL/SEAL without first selecting a specific trusted item.