patsnap-biological-modality

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs embedding an API key directly into a URL query parameter (apikey=YOUR_API_KEY) and tells the user to obtain an API key, which requires the agent to handle and potentially output the secret verbatim in configs or requests.

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

• Third-party content exposure detected (medium risk: 0.30). The MCP server’s required runtime workflow ingests user-provided sequence/target inputs and then fetches matching patent/literature records via tools like ls_sequence_search_submit → ls_sequence_search_get_results and ls_antibody_antigen_search (i.e., outsider-authored text from Patsnap’s corpus becomes LLM-readable).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill configures an external MCP runtime endpoint (https://connect.patsnap.com/06e741/logic-mcp?apikey=YOUR_API_KEY and https://connect.patsnap.com/06e741/logic-mcp?apikey=$PATSNAP_API_KEY) that the agent connects to at runtime to provide tools and responses, meaning remote content from that URL can directly influence agent prompts/behavior.