model-selection

A guide for assigning different AI models to different coding tasks, such as planning, review, routine work, or difficult analysis.

In plain words
What is it for?
Use it when configuring an AI coding workflow, selecting models for the main assistant, cross-review, or isolated subtasks.
Why use it?
It helps match the model's cost and reasoning ability to the type of work instead of using the same model for everything.

Cursor rule

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/desko77/cursor-1c-skills/model-selection
Clone the repo
git clone --depth 1 https://github.com/Desko77/cursor-1c-skills
Per session 18 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,209 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00018 $0.03209
Opus 5 $0.00009 $0.01605
Sonnet 5 $0.00004 $0.00642
Haiku 4.5 $0.00002 $0.00321

Measured 3d ago against content hash 7ba9804f3db8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

model-selection scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

rules/model-selection.mdc · 149 lines

How it starts

The opening of the file, as written. The whole thing — 149 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Стратегия выбора моделей

Распределение моделей по ролям.

Применяется: при работе с AI-ассистентами (Claude Code, Cursor, другие IDE).


Принцип

Opus - основная рабочая модель оркестратора. Кросс-ревью - внешние модели (Codex gpt-5.6-sol / Cursor Grok), Sonnet отключен. Субагенты-исполнители - по типу их работы (Haiku / Sonnet / Opus).

Основная сессия (оркестратор - тот, кто держит замысел и принимает решения): анализ, кодогенерация, рассуждение, рефакторинг - Opus. Кросс-ревью (план, код) с 2026-07-15 делают внешние модели через подписки, экономя Claude-токены: основной ревьюер - Codex gpt-5.6-sol (нативный codex review), опционально Grok через Cursor; Sonnet из ревью отключен (детали и fallback - в промптах кросс-ревью). Субагентам, которым делегируется ИЗОЛИРОВАННАЯ работа (разведка, механическая выборка, отдельный анализ), модель подбирается по типу задачи - от Haiku на механике до Opus на тяжелом анализе (см. "Модель субагента по типу задачи").

Третий уровень - Fable 5 (claude-fable-5) точечно поверх Opus на long-horizon автономных прогонах и самых тяжелых задачах (см. раздел ниже). Не дефолт: дороже Opus вдвое и быстрее расходует недельную квоту, включается под конкретный заход через /model.


Распределение

Opus - все задачи

Категория Примеры
Архитектура Проектирование, выбор технологий, breaking changes, миграция
Безопасность Ревью кода, аудит RLS, поиск уязвимостей
Production-баги Порча данных, race conditions, регрессии
Кодогенерация Формы, тесты, запросы, скрипты, конфигурации
Рефакторинг Упрощение, извлечение методов, консолидация
Исследование Навигация по коду, поиск, документация
Итеративная работа Баг-фиксы, реализация фич, тестирование

Кросс-ревью - внешние модели (Codex / Grok), не Sonnet

Когда Кто ревьюит
После завершения кода Codex gpt-5.6-sol (codex review, сам читает git diff) + опц. Grok через Cursor
После составления плана Codex gpt-5.6-sol (нативный codex review) + опц. Grok

Read the full file on GitHub · 149 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 149 lines · 18 tokens per session scan A 7ba9804f3db8

Subscribe to this mod's changes

model-selection is a cursor rule published in the GitHub repository Desko77/cursor-1c-skills (53 stars, last pushed 5d ago), licensed MIT. It adds 18 tokens to every session and 3,209 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other cursor rules, from other repositories

121-java-unit-testing

Java Unit testing guidelines.

alkoleft/platform-context-exporter · 0 tokens

architectural-planning

Cursor rule "architectural-planning" from alkoleft/platform-context-exporter, covering architectural planning for level 4 tasks, 🔍 architectural planning overview, 📋 architectural planning principles, 📋 architectural requirements analysis and functional requirements analysis.

alkoleft/platform-context-exporter · 6,209 tokens

reflection-comprehensive

Cursor rule "reflection-comprehensive" from alkoleft/platform-context-exporter, covering comprehensive reflection for level 4 tasks, 🔍 comprehensive reflection overview, 📋 reflection template structure, 1. system overview and system overview.

alkoleft/platform-context-exporter · 6,108 tokens

workflow-level4

Cursor rule "workflow-level4" from alkoleft/platform-context-exporter, covering comprehensive workflow for level 4 tasks, 🔍 level 4 workflow overview, 🔄 level transition handling, 📋 workflow phases and phase 1: initialization.

alkoleft/platform-context-exporter · 3,729 tokens

creative-phase-uiux

UI/UX Design Guidelines and Process for the Creative Phase.

alkoleft/platform-context-exporter · 0 tokens

optimization-integration

Cursor rule "optimization-integration" from alkoleft/platform-context-exporter, covering memory bank optimization integration, 🔄 optimization integration flow, 📋 optimization component registry, 🔄 optimization initialization sequence and 🔍 optimization configuration.

alkoleft/platform-context-exporter · 2,319 tokens