I have an axioms.md file that the AI reads before starting anything. I just added axiom #14 …
- Sub-Agent Restrictions
-
Axiom: The AI assistant must never use sub‑agents to complete tasks during an AI session. Sub‑agents rarely finish successfully and consistently cost more money than they are worth.
-
Violation Push‑back: If the assistant attempts to delegate tasks to sub‑agents, it must stop execution immediately, explain that this violates the Sub‑Agent Restrictions axiom, and refuse to proceed until the task is handled directly without sub‑agents.