alethe
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local AI agent CLI tools (e.g.,
claude,codex,agy,kimi) via a Python orchestrator (scripts/orchestrate.py) or direct shell commands. This command execution is the primary intended functionality required to delegate and supervise work waves. - [INDIRECT_PROMPT_INJECTION]: The skill ingests arbitrary user tasks and interpolates them into briefs for sub-agents, creating a surface for indirect prompt injection. This risk is heavily mitigated by instructions that enforce a "worker contract," which explicitly forbids workers from fanning out to other agents, denies access to sensitive tools (like push/commit), and restricts writers to isolated git worktrees.
Audit Metadata