Model Delegation

Initiate the orchestrator-worker cost model for this session. The main-loop model (expensive, metered) orchestrates, specs, verifies, and validates only; ALL drafting and mechanical work (code, docs, tests, sweeps, research) runs in Opus 4.8 subagents billed to plan allowance. Invoke at session start or whenever token cost matters.

14 starsAI, LLMs & Agents

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add erhantezakar/claude-fable-5-spanking-claude-opus-4.8 --skill model-delegation
Manual install
git clone --depth 1 https://github.com/erhantezakar/claude-fable-5-spanking-claude-opus-4.8.git /tmp/claude-fable-5-spanking-claude-opus-4.8
mkdir -p ~/.claude/skills
cp -r /tmp/claude-fable-5-spanking-claude-opus-4.8/. ~/.claude/skills/model-delegation

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Model delegation (orchestrator thinks, workers write)

Purpose: when the main-loop model bills to metered usage credits and
subagent models bill to a subscription plan, the expensive model must never
do the writing. Output tokens are the costly side; drafting is output-heavy;
verification is read-heavy. So the orchestrator reads and decides, the
workers write.

On invocation, adopt this operating mode for the rest of the session.

## Model policy

| model | role | billing |
|---|---|---|
| main loop (e.g. Fable 5) | decompose, spec, review, validate, decide, talk to the user | metered credits, minimize output |
| opus-4.8 subagents | write code/docs/tests, mechanical sweeps, research, exploration | plan allowance |
| sonnet / haiku | never, regardless of task size | - |

These are defaults with one standing override: if a worker's output does not
meet the bar after one revision round, the orchestrator redoes that piece
itself. Judge the output, not the price tag; escalating costs less than
shipping mediocre work.

## Mechanics

- Spawn workers via the Agent tool with `model: "opus"`. Choose
  `subagent_type` per task: `general-purpose` for implementation, `Explore`
  for read-only search, `Plan` for design drafts.
- Independent tasks launch in ONE message so they run in parallel.
- Workers mutating files in parallel get `isolation: "worktree"`.
- Continue an existing worker with SendMessage instead of re-spawning; a
  fresh Agent call starts with zero context.
Read the full skill on GitHub

Details

Skill name
model-delegation
Repository
erhantezakar/claude-fable-5-spanking-claude-opus-4.8
License
MIT
Repo updated
17 July 2026

FAQ

What does the Model Delegation skill do?

Initiate the orchestrator-worker cost model for this session. The main-loop model (expensive, metered) orchestrates, specs, verifies, and validates only; ALL drafting and mechanical work (code, docs, tests, sweeps, research) runs in Opus 4.8 subagents billed to plan allowance. Invoke at session start or whenever token cost matters.

How do I install the Model Delegation skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add erhantezakar/claude-fable-5-spanking-claude-opus-4.8 --skill model-delegation. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Model Delegation popular?

It lives in erhantezakar/claude-fable-5-spanking-claude-opus-4.8, which has 14 GitHub stars and 0 forks.

Related skills

Openrouter Integration

14

bnishit/openrouter-skill

Connect apps to hundreds of AI models through OpenRouter — live model and discount discovery, batch and service-tier cost controls, multimodal chat, exact cost and credit diagnostics, provider routing, reasoning, tool calling, structured output validation, starter templates, production playbooks, and verification scripts. Use when an agent needs to add, compare, audit, or debug OpenRouter models, pricing, promotions, credits, routing, requests, or generated assets.

AI, LLMs & Agents

Humanizer

14

hannsxpeter/humanizer

Humanize and de-slop AI-sounding prose while preserving meaning, specific detail, and genuine human quirks. When a voice sample or style profile is available, rewrite in that writer's actual voice. Remove recurring AI tells such as inflated significance, stock vocabulary, uniform rhythm, excessive hedging, formulaic structure, and formatting artifacts. Also clean suspicious invisible Unicode and copy-paste residue conservatively, without treating them as proof of authorship. Use whenever a user wants to humanize, de-AI, de-slop, or de-robotify text; fix prose that sounds corporate, generic, li

AI, LLMs & Agents

Yandex Direct Manager

14

ai-hub-open/yandex-direct-manager

Тактический пайплайн создания рекламной кампании в Яндекс.Директе через MCP — сбор семантики через Wordstat, кластеризация на НЧ/СЧ/ВЧ и Поиск/РСЯ, минусация (3 уровня + кросс), создание объявлений по лимитам Директа, аудит счётчика и настройка целей в Метрике через MCP, выбор стратегии торгов и заливка в DRAFT. Используй когда пользователь говорит «сделай РК в Директе», «собери семантику», «нужны ключи и минусы», «напиши объявления», «запусти кампанию», «залей через MCP». Умеет выполнять отдельный этап — только семантика, только минусы, только объявления — без полного брифа. Не для аналитики,

AI, LLMs & Agents

Yocto Agent Skills

13

Higangssh/yocto-agent-skills

Official-doc-first Yocto Project and BitBake agent skills for creating or reviewing recipes, layers, images, machine/distro config, bbappends, build logs, QA errors, fetch/patch/package/rootfs failures, and modern BitBake override syntax. Use when working with Yocto/OpenEmbedded files such as .bb, .bbappend, .bbclass, layer.conf, local.conf, bblayers.conf, machine config, distro config, image recipes, kernel recipes, or BitBake command output.

AI, LLMs & Agents

Goog Geo

11

vishalmdi/goog-geo

Audits any website URL for Google AI Search readiness based on Google's official AI optimization guide. Use when the user wants to know if their page is eligible for Google AI Overviews or AI Mode, wants a GEO score, wants to optimize for AI-powered search, or wants to know why they're not being cited in AI search results. Scans the live URL using playwright-cli (auto-installs if needed). Trigger phrases: 'GEO audit', 'AI search optimization audit', 'AI overview optimization', 'generative engine optimization check', 'optimize for ChatGPT/Perplexity', 'AI readiness check'. For content strategy

AI, LLMs & Agents

Workbuddy Dual Edition

13

ZenHG/workbuddy-dual-edition

处理腾讯 WorkBuddy 国内版(workbuddy.cn)与国际版(workbuddy.ai)在同一台电脑上的共存问题。触发词：双开、同时开两个、开第二个没反应、装国内版却变国际版、只能登录国际版、重启后被覆盖、卸载重装还复发、两版互相覆盖、国际版登不上、国内版登不上、WorkBuddy 两版冲突、WorkBuddyAI.exe 突然出现。

AI, LLMs & Agents

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.