small-molecule-rd-analysis
Fail
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: A hardcoded API key was detected in the tool configuration URL within SKILL.md. \n
- Evidence: The URL
https://connect.zhihuiya.com/713886/logic-mcp?apikey=sk-3F0NsY7KOt2ZyO72XkgwUIwD80xSfBjCNKp7juA92d0HWpKucontains an exposed secret key. While the domain belongs to the skill's vendor (PatSnap), hardcoding secret keys in instruction files is an unsafe credential management practice.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data to use as tool parameters without explicit boundary markers or sanitization, creating a potential attack surface.\n - Ingestion points: User-provided SMILES, InChI, CAS numbers, and compound names as described in the workflow section of
SKILL.md.\n - Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the processing workflow.\n
- Capability inventory: The skill invokes external MCP tools to query chemical databases based on user input.\n
- Sanitization: The instructions do not specify any validation, escaping, or filtering for the external identifiers before they are passed to the MCP tools.
Recommendations
- AI detected serious security threats
Audit Metadata