search-cold-skills

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

• Third-party content exposure detected (high risk: 0.75). At runtime, scripts/query.sh regenerates ~/.agents/skills-cold/INDEX.md by reading outsider-authored SKILL.md files from cold-storage directories (which may contain third-party skill text), and then the agent uses the resulting matches to read those SKILL.md bodies into LLM context (indirect prompt injection risk).