claude-handoff
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the
claude --bgCLI tool to launch new background processes.\n- [DYNAMIC_EXECUTION]: The agent is directed to dynamically construct a CLI command at runtime using strings generated from conversation content and user arguments.\n- [INDIRECT_PROMPT_INJECTION]: The skill uses potentially untrusted data from conversation history and user input to form the command and prompt for the next agent.\n - Ingestion points: User arguments and conversation history are processed in SKILL.md.\n
- Boundary markers: There are no explicit delimiters or boundary markers defined to separate user data from the command structure.\n
- Capability inventory: The skill utilizes shell command execution via the
claudeCLI.\n - Sanitization: While the skill advises redacting sensitive information like API keys, it lacks specific instructions for sanitizing shell metacharacters that could lead to command injection.
Audit Metadata