claude-handoff

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute the claude CLI with the --bg flag to launch a background process. This involves shell command execution to initiate new agent sessions.
  • [INDIRECT_PROMPT_INJECTION]: The skill seeds a new agent session with a summary of the current conversation and external references, creating a potential vector for malicious instructions to persist across sessions.
  • Ingestion points: Uses the existing conversation history, user-provided arguments, and referenced file paths or URLs as input for the handoff summary.
  • Boundary markers: Employs shell quoting in the launch command but provides no specific delimitation (e.g., XML tags) for the content within the summary itself to prevent instruction obedience by the next agent.
  • Capability inventory: The skill performs file system writes to a temporary directory and executes shell commands via the claude CLI.
  • Sanitization: Includes an explicit safety directive to redact sensitive information, such as API keys and passwords, from the summary before it is passed to the next session.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 09:28 AM