beastmode-hermeslisted

# Beastmode Hermes - Hermes Harness Adapter This skill is **only the harness mechanics**. The framework, tier-routing rule, verifier-first design principle, and self-improvement loop live in the canonical `beastmode` skill (v2.3.0) - load it first and follow it. This adapter tells you which Hermes primitives implement which seat and how to wire them together. ## Only the harness mechanics | Universal seat | Hermes implementation | |---|---| | Director / Lead (frontier tier) | The Hermes session itself - frontier model | | Watcher (adversarial review, frontier) | A second `delegate_task` on a frontier model only after the user explicitly names it | | Executor (economy tier) | `delegate_task` children pinned to MiniMax-M3 | | Loop engine (continues until done) | The agent's own phase loop (director prompts itself between subagent results) | | Anti-spin circuit breaker | Director judgment + the run's usage / wall-clock budget; abort on 3 identical blockers | | Worker-contract enforcer | `~/.hermes/config.yaml` `permissions` / `toolsets` denials + a starter worker contract in the prompt | | Live progress visibility | `~/.hermes/cache/delegation/live/<id>/task-<n>.log` (tail in a terminal) | | Remote supervision | Optional: any deliver='telegram' cron mirroring, or the parent OpenClaw channel | | Durable goal record | Optional: write the acceptance contract to `<repo>/GOAL.md` (universal fallback) | | Self-improvement log | `<repo>/.learnings/BEASTMODE.md` (universal) | Automa