find-examples

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

• Third-party content exposure detected (low risk: 0.10). The runtime workflow’s “Find the matching example” step conditionally reads outsider-authored free text by fetching GitHub tree/contents for playcanvas/engine examples (gh api ... git/trees/... and gh api .../contents/...) and decoding the returned .example.mjs/example files when local examples are absent.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill invokes gh api to fetch raw example source from GitHub at runtime (e.g., repos/playcanvas/engine/contents/examples/src/examples//.example.mjs?ref=$TAG), which downloads remote code/content that would be read into the agent runtime and could be injected or executed.