pixi

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs a runtime fallback to WebFetch https://pixijs.download/release/docs/llms.txt (public web content) and then use its linked .html.md pages, which would be outsider-authored free text ingested into the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). SKILL.md explicitly instructs the agent to WebFetch https://pixijs.download/release/docs/llms.txt at runtime, and that fetched file (and the linked .html.md pages) would be incorporated into the agent's context to control prompts/instructions.