handoff
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing potentially untrusted conversation history and artifacts.
- Ingestion points: Conversation history and referenced external artifacts (file paths or URLs).
- Boundary markers: Absent; there are no instructions to treat ingested data as data rather than instructions.
- Capability inventory: File system write access to the OS temporary directory.
- Sanitization: The instructions include a manual redaction step for sensitive data ('Redact any sensitive information'), which is a security best practice, but no automated sanitization or instruction to ignore embedded commands.
- [COMMAND_EXECUTION]: The skill directs the agent to interact with the file system outside of the defined project workspace.
- Evidence: 'Save to the temporary directory of the user's OS
- not the current workspace.'
Audit Metadata