mmf

Route coding tasks across planning, implementation, and review models with automated test gates.

Updated Jun 17, 2026
One-click install
npx skills add https://github.com/Adam-Luciano-MDB/multi-model-flow --skill mmf
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mmf
Source: https://github.com/Adam-Luciano-MDB/multi-model-flow/tree/main/skills/mmf
Command: npx skills add https://github.com/Adam-Luciano-MDB/multi-model-flow --skill mmf

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This skill solves the inefficiency of using expensive, high-tier AI models for every coding task by implementing a tiered routing system that balances cost, speed, and quality.

Core Features & Use Cases

  • Tiered Model Routing: Automatically routes planning to Opus, implementation to Haiku or local Ollama models, and reviews to Sonnet.
  • Automated Recovery: Features a hard test gate and a tiered fix loop that attempts to resolve bugs using cheaper models before escalating to a full re-plan.
  • Use Case: Use this in a large-scale refactoring project to ensure high-stakes architectural decisions are handled by Opus while routine implementation and testing are offloaded to local or cost-effective cloud models.

Quick Start

Invoke the multi-model-flow skill by typing /mmf followed by your task description and any desired flags like [auto] or [openrouter].

Frequently Asked Questions about mmf

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I optimize LLM API costs for large-scale coding workflows?

You can optimize LLM API costs for coding workflows by using a tiered routing system that delegates architectural planning to high-reasoning models and routine implementation to cost-effective local or cloud models.

Can I use local Ollama models for AI code generation and testing?

Yes, you can use local Ollama models for AI code generation by integrating them into a multi-phase pipeline that offloads routine implementation and automated testing from expensive cloud LLMs.

What is automated model routing in a multi-phase coding pipeline?

Automated model routing in a coding pipeline assigns high-stakes planning to high-tier LLMs, implementation to cheaper models, and reviews to mid-tier models, balancing development speed, cost, and code quality.

How do automated test gates and recovery loops handle LLM generated code?

Automated test gates block unverified LLM generated code, while tiered recovery loops attempt to resolve bugs using cheaper models before escalating to a full architectural re-plan.

Does multi-model routing work for complex software refactoring tasks?

Multi-model routing works for complex software refactoring by ensuring high-stakes architectural decisions are handled by advanced reasoning models while offloading routine implementation to cost-effective endpoints.

When should I avoid using a tiered multi-model coding pipeline?

You should avoid a tiered multi-model coding pipeline for simple, single-file edits where the overhead of routing between local Ollama instances and cloud LLMs outweighs the cost savings.