openspec-reverse-engineer

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s workflow ingests and extracts text from outsider-authored sources during “Explore Codebase”/“Extraction Phase” (e.g., scanning and converting SKILL.md files found in repo directories like .github/skills/, .agents/skills/, .claude/skills/), where those SKILL.md contents may be authored by parties other than the operating user and become LLM-readable prose in the agent context.