handoff
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill summarizes arbitrary conversation data, creating a surface for indirect prompt injection where instructions embedded in the dialogue may be carried over to future sessions.
- Ingestion points: Conversation history processed for summarization in SKILL.md.
- Boundary markers: The skill does not define delimiters or warnings to treat the conversation content as untrusted data.
- Capability inventory: The skill implies the ability to write files to the system's temporary directory (SKILL.md).
- Sanitization: It explicitly instructs the agent to redact credentials and PII, which mitigates data exposure but does not prevent instruction injection.
- [DATA_EXFILTRATION]: Writing conversation summaries to the OS temporary directory could result in local data exposure if the directory permissions are not restrictive on a multi-user system.
- [NO_CODE]: The skill consists of instructional text and metadata with no executable code or scripts provided.
Audit Metadata