memory-ingest

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires preserving and writing source content verbatim into notes (e.g., "Preserve all source content verbatim"), which would force the LLM to include any secrets present in the input in its generated output, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.75). The required workflow ingests “pasted documents, email threads, meeting transcripts, conversation logs” (user-provided free text) and then writes a “Create the Source Note” containing verbatim content into the LLM context; if that pasted text includes outsider-authored material, it becomes outsider free text in the agent’s context via the raw-input parsing → source-note creation path.