otio-docs

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow fetches and text-extracts documentation from public web sources (e.g., https://opentimelineio.readthedocs.io/... and raw GitHub READMEs) via urllib.request.urlopen in fetch_url() and then ingests the extracted prose into the agent context through fetch/search/section outputs.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill fetches documentation at runtime from external URLs (e.g., https://opentimelineio.readthedocs.io/... and raw GitHub files at https://raw.githubusercontent.com/AcademySoftwareFoundation/OpenTimelineIO/main and https://raw.githubusercontent.com/OpenTimelineIO/OpenTimelineIO-Plugins/main) and injects that remote content into its responses, so those runtime-fetched pages can directly control the agent's output/context.