llm-ai-security

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly includes curl examples that interpolate and send a bearer token/session cookie in headers and contains payloads that instruct the model to "output the admin password" or exfiltrate system prompt data, which require secrets to appear verbatim in requests/responses.

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

• Third-party content exposure detected (high risk: 1.00). This skill's workflow explicitly instructs sending and processing untrusted, user-controlled content (e.g., "document uploads, pasted text, web page summaries, database records, emails, or calendar events" in Step 4, plus web_search/tool-invocation tests in Steps 6 and 9), so the agent is expected to ingest and act on arbitrary third‑party content that could carry indirect prompt injections.