delegatelisted

# Delegate Run before every `Agent()` call — cheap enough to reload each time. Spawning is the dominant token cost: every subagent rebuilds its context cache from scratch, and cache-creation, not model output, is the bulk of the bill. The cheapest work is the spawn that never happens. ## Pre-Delegation Checklist 1. **Total scope size** — sum the estimated diff size (excluding comments) across the whole batch, not per item. Under ~300 lines total (soft guesstimate): execute directly in the coordinator, or fold into an existing/sibling agent via `SendMessage` if one is already live in scope; do not spawn a new agent. 2. **Genuine parallelism** — is there a real wall-clock or file-independence need, or would sequential work merely take "a bit longer"? No real need → one agent, sequential. 3. **Reuse** — is an agent already live in the same file/domain scope? → `SendMessage` it. Accumulated context beats a cold spawn (see § Agent Reuse in `grand-admiral`). 4. **Model tier** — set explicitly on this spawn per the table below. Never leave it to the agent's frontmatter fallback. 5. **Worktree** — for code-mutating agents: pre-create it and inject the resolved SHA. Never rely on `isolation: "worktree"` alone (see § Worktree Isolation in `grand-admiral`). 6. **Monitoring** — is a watchdog running for this session (MCP preferred, else the built-in Monitor)? An un-monitored dispatch is a doctrine violation (see § Recovery in `grand-admiral`). 7. **Development work?** — brief the goal