claude-handoff
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill summarizes the current conversation to create the starting prompt for a new background agent, which could allow malicious instructions from the conversation to influence the new agent's behavior.\n
- Ingestion points: The handoff summary is derived from the current conversation history and user-supplied arguments as described in
SKILL.md.\n - Boundary markers: No explicit delimiters or instructions are provided to the secondary agent to treat the handoff summary as untrusted data.\n
- Capability inventory: The skill utilizes the
claudeCLI to launch background processes.\n - Sanitization: The instructions correctly mandate redacting credentials (API keys, passwords), but do not include checks for prompt injection content.\n- [COMMAND_EXECUTION]: The skill directs the agent to execute a shell command to initialize background sessions.\n
- Evidence: The use of the
claude --bg --name "<descriptive name>" "<handoff summary>"command inSKILL.md.
Audit Metadata