skill-optimizer

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md’s runtime workflow for skill optimization ingests user-authored benchmark tasks from skills/<name>/skillopt-benchmark.jsonl (or generated bootstrap content) as the task/evaluation prompts during the forward pass, so an outsider can supply free text that the LLM reads without selecting a specific item first.