claude-handoff
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a command-line tool
claude --bgto initiate background processes. This relies on the local environment having theclaudeCLI installed and authorized. - [INDIRECT_PROMPT_INJECTION]: The skill uses conversation history to generate a summary that serves as the system prompt for a new agent session. This creates an attack surface where instructions embedded in the original conversation data could potentially influence the behavior of the subsequent background agent.
- Ingestion points: Conversation history and user arguments are ingested in
SKILL.mdto create the handoff summary. - Boundary markers: The summary is passed as a quoted string in the shell command
claude --bg --name "<name>" "<handoff summary>", providing basic shell delimiter protection. - Capability inventory: The skill has the capability to execute shell commands via the
claudeCLI. - Sanitization: The skill explicitly instructs the agent to redact sensitive information like API keys, passwords, and PII before generating the prompt for the next session.
Audit Metadata