project-development

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s provided runtime workflow is the staged LLM pipeline template (scripts/pipeline_template.py) where the Process stage reads arbitrary per-item prompt.md (derived from acquire) and sends its free-text contents into the LLM, enabling outsider-authored prompt injection if acquire pulls from an outsider-controlled feed/queue without preselecting a specific trusted item.