weapon-forge

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s required runtime workflow explicitly has the agent open and read ai-tools/command-briefs/<guardian-name>-command-brief.md end-to-end and ai-tools/skills/<weapon-name>/research/research-summary.md/index.md, which are outsider-authored free text inputs (produced by command-center and loremaster) that get synthesized into the LLM’s context during forging.