service-design-ingest

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill instructs the agent to "ingest, don't summarize" source material and to state "current text" and "proposed text" (and optionally save sources), which can require verbatim reproduction of any secrets present in pasted files/logs/code without any redaction guidance.

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts "pasted content, file paths, or URLs" and requires the agent to "Read the source material in full — ingest, don't summarize" (SKILL.md), so it can ingest untrusted/user-generated third‑party content via URLs that the agent will interpret and use to drive artifact updates.