distill

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly forwards "all arguments directly" to /doc2qra, which requires the LLM to include user-supplied argument values verbatim (and those may contain API keys or secrets), creating an exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). The distill compatibility shim simply forwards user-provided --text/--url/--file arguments to the /doc2qra workflow (via exec bash "$DOC2QRA_RUN" "$@"), meaning outsider-authored free text can be ingested into the required runtime when provided through this interface.