Delegation

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local utility script (~/.claude/skills/Agents/Tools/ComposeAgent.ts) using bun run. This command is used to generate sub-agent identities and prompts based on user-provided traits and tasks.
  • [SAFE]: The skill promotes secure execution patterns, such as the use of isolation: "worktree" for parallel agents, which prevents conflicting file modifications and ensures a clean environment for sub-tasks.
  • [PROMPT_INJECTION]: The skill identifies a potential indirect prompt injection surface (Category 8) through its delegation mechanism.
  • Ingestion points: User-supplied task descriptions, traits, and prompts are used as arguments for the Task() tool and the identity composition script.
  • Boundary markers: Absent; the examples do not demonstrate the use of delimiters or 'ignore' instructions to wrap untrusted data within the delegated prompts.
  • Capability inventory: The delegated sub-agents (e.g., Engineer, Architect) typically have access to powerful tools including file system operations and shell command execution.
  • Sanitization: There is no evidence of sanitization or validation of the user input before it is interpolated into sub-agent prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 01:03 AM