Route user prompts to AI model tiers with cascading classifier fallbacks.
npx skills add https://github.com/alinaqi/maggy --skill model-routing-system
Please help me install this Agent Skill. Skill: Model Routing System Source: https://github.com/alinaqi/maggy/tree/main/skills/model-routing Command: npx skills add https://github.com/alinaqi/maggy --skill model-routing-system
It solves the problem of getting consistently good, cost-aware answers by automatically classifying each user prompt and delegating it to an appropriate model with cascading fallbacks and verification when the classifier or tools fail.
Ask Claude Code to handle your request normally and let the route-task hook classify the prompt, delegate to the selected tiered model, and fall back automatically if routing or tools fail.
High-intent search queries and answers about installing and using this skill.
Multi-model routing classifies each prompt by cost and complexity, delegating it to the appropriate AI model tier. This ensures cost-aware inference by using cheaper local classification first and escalating to higher-quality models only when needed.
You route prompts to different AI models in Claude Code by submitting your request normally. A hook automatically classifies the prompt through a 9-tier pipeline, delegates it to the selected model, and applies cascading fallbacks if routing or tools fail.
Yes, the model routing system supports fallback recovery after context compaction. It uses a routing cache to maintain routing continuity and retries classification across multiple backends if local tools like Ollama become temporarily unavailable.
Yes, you can use specific models for tasks like security architecture and bulk extraction. The system enforces task-type overrides, routing security and architecture prompts to Claude while delegating bulk extraction tasks to Gemini Flash-Lite.
When the prompt classifier fails during multi-model routing, the system executes a tool fallback protocol. It swaps to external research backends and can auto-evaluate results to select the better-performing option, ensuring reliable output.
alinaqiRoutes user prompts to tiers via UserPromptSubmit hook and delegation scripts.
AytsuuSuggest manual context compaction at strategic workflow intervals.
artimathRoute Claude Code extension build requests to appropriate builders for skills, hooks, plugins, marketplaces, slash commands, and agents.
planetaryescapeRoute user messages to optimal AI models via task classification and cost-aware tier selection.
