space-earth-science-explorer

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

• Third-party content exposure detected (high risk: 0.75). At runtime, the required action query_space_science_data searches NASA/NOAA/USGS public APIs and returns results whose fields (e.g., titles/descriptions/landing-page metadata) are derived from outsider-authored web/API content, which the agent then ingests into the LLM context as structured text.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs runtime calls to the AgentPMT MCP API to fetch live schema and instructions (e.g., https://api.agentpmt.com/mcp/), and those fetched instructions are treated as authoritative and can directly control the agent's prompts/behavior.