earth2studio-discover

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 live documentation pages from nvidia.github.io/earth2studio/... (public web content), which is outsider-authored text that can be ingested into the agent’s LLM context for model/data-source discovery.

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

• Potentially malicious external URL detected (high risk: 0.80). This skill explicitly requires fetching live documentation at runtime (e.g. https://nvidia.github.io/earth2studio/modules/models_px.html, the other nvidia.github.io earth2studio module pages, and the lexicon on GitHub https://github.com/NVIDIA/earth2studio/tree/main/earth2studio/lexicon), and those fetched pages are used to determine model/data recommendations and lexicon compatibility—i.e., external content directly controls the agent's outputs.