ai-model-orchestrator

根据任务复杂度、成本与延迟，动态选择最优的AI模型。将请求路由至Gemini、GPT-4、Claude，或本地LLM，以最大化效率。

SKILL.md

--- frontmatter

name: ai-model-orchestrator
description: Dynamically selects the optimal AI model based on task complexity, cost, and latency. Routes requests to Gemini, GPT-4, Claude, or local LLMs to maximize efficiency.
status: implemented

AI Model Orchestrator

This skill ensures that the most appropriate and cost-effective intelligence is used for every task.

Capabilities

1. Task Triage

•Analyzes the "Hardness" of a prompt or code task.
•Routes simple tasks (e.g., translation, basic cleanup) to faster/cheaper models.
•Routes complex tasks (e.g., architectural design, deep debugging) to state-of-the-art models.

2. Multi-Provider Fallback

•Automatically switches providers if an API is down or throttled.
•Compares outputs from multiple models for critical tasks (consensus building).

Usage

•"Execute the code review using the most cost-effective model that maintains high precision."
•"Orchestrate a multi-model evaluation of our new security architecture."

Knowledge Protocol

•This skill adheres to the knowledge/orchestration/knowledge-protocol.md. It automatically integrates Public, Confidential (Company/Client), and Personal knowledge tiers, prioritizing the most specific secrets while ensuring no leaks to public outputs.