handoff
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for file management and clipboard access without requiring a confirmation gate.- [INDIRECT_PROMPT_INJECTION]: The skill facilitates transferring context between independent agent sessions, creating a potential path for persisting malicious instructions from one session to the next.
- Ingestion points: The entire conversation history, which may include untrusted data from files or tool outputs, is processed to generate the handoff.
- Boundary markers: The instructions define a strict three-section format and require the agent to strip backstory and project assumptions.
- Capability inventory: The skill uses shell commands to write to the
/tmp/directory and access the system clipboard viapbcopyorxclip. - Sanitization: The instructions explicitly forbid pasting file contents, diffs, or command output, and impose a word count limit to minimize the injection surface.
Audit Metadata