advisor-modelisted
Install: claude install-skill cerebrocybersolutions/libro
# Advisor Mode
### Tiered Model Routing | Plan-First | Budget-Controlled
> **V1 scope:** This skill is a **wrapper / classification-and-dispatch harness**, not a native model runtime. It selects the tier (A+/A/B/C) and the venue (cloud Anthropic API vs local Ollama) and hands off; it does not own inference. Treat advisor-mode as the **HR layer** in the executive-stack org chart below — it routes, it doesn't think for you.
This skill routes tasks to the correct model tier and enforces a plan-before-act discipline.
Simple tasks stay cheap. Strategic tasks get Opus-level reasoning where it counts.
**Core principle:** Opus advises on the PLAN. The executor carries out the plan.
Never use Opus to execute when it could plan instead.
---
## EXECUTIVE STACK — Where advisor-mode fits
advisor-mode is one of **three** orchestration skills, each owning a different layer.
They are NOT alternatives — they cover different axes and are usually composed, not swapped:
| Org-chart role | Skill | Axis |
|---|---|---|
| HR / routing (pick the rank for the task) | **advisor-mode** (this skill) | Classification + tier selection |
| Parallel specialist advisors (surface disagreement) | **council-mode** | Parallel adversarial comparison |
| Executor + final arbiter (run the chain, gate, escalate) | **orchestrator-mode** | Sequential gate-and-escalate |
Claude tier mapping (canonical): A+ = CEO / Chief Strategist (Opus solo) · A = Director with
Advisor (Sonnet + Opus advisor tool) · B = Direct