night-park

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires reading CLAUDE_CODE_SESSION_ID and embedding the full UUID verbatim into disk files, engram saves, and the final printed resume command, forcing the LLM to output a secret-like value.

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

• Third-party content exposure detected (medium risk: 0.65). The skill drafts the handoff by “Read the conversation context” (Step 3) and then writes that derived text into the LLM/agent context via mem_save/mem_session_summary, so any outsider-authored messages present in the conversation history can be ingested into the model’s context through the conversation-to-summary path (indirect prompt injection risk).

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)