rowan

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill contains explicit examples that assign API keys in code, export API keys in shell commands, and print/manage webhook secrets (e.g., printing secret.secret and storing webhook_secret), which requires the agent to handle and potentially output secret values verbatim — an exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes Rowan workflow submission and result retrieval using user-provided molecule strings/sequences and webhook payloads, meaning the agent at runtime can ingest outsider-authored free text supplied by the caller (e.g., SMILES/SMARTS/InChI and sequences) without any need to select a specific pre-existing item first.