handoff
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes conversation history (untrusted data) to generate a summary for a future agent session. This creates a multi-step attack surface where instructions embedded in the conversation history could be included in the handoff document, potentially influencing the behavior of the next agent.
- Ingestion points: Current conversation history and user-supplied arguments are used to generate the summary.
- Boundary markers: The instructions do not specify specific delimiters for the summarized content to separate it from agent-specific instructions.
- Capability inventory: The skill performs a file write to the OS temporary directory.
- Sanitization: The skill includes explicit instructions to redact sensitive data like API keys and passwords, which mitigates data exposure but not necessarily instruction injection.
- [COMMAND_EXECUTION]: The skill instructs the agent to write a file to the system's temporary directory (
/tmp,$TMPDIR, or%TEMP%), involving file system operations outside the current project workspace.
Audit Metadata