Skill

which-ai

Routes AI tasks to optimal LLMs by analyzing budget, deployment (local/cloud), and modality (text/vision/coding). Fetches live model data via curl and runs Python router script.

Python

Bash

ai-ml

npx claudepluginhub roeytidhar/which-ai-skills --plugin which-ai

Tool Access

This skill uses the workspace's default tool permissions.

Preview

You are the `which-ai` orchestrator.

SKILL.md

Similar Skills

route

Routes tasks to optimal LLMs by auto-classifying type (research, generate, analyze, code, query, image) and complexity (budget, balanced, premium) via heuristics, Ollama, or cheap APIs for cost savings.

llm-router

model-router

586

Routes tasks to optimal AI models across Anthropic, OpenAI, Gemini, Moonshot, Z.ai, GLM providers based on task type, complexity, cost. Features setup wizard, classifier, secure API keys. Activates on model switch or optimization requests.

4 files

sundial-org-awesome-openclaw-skills-4

openrouter-model-routing

1.9k

Routes OpenRouter API calls to optimal models by task (e.g., code review to Claude-3.5-Sonnet) or prompt complexity for cost, quality, latency optimization in multi-model apps.

7 files5 tools

openrouter-pack

Stats

Parent Repo Stars12

Parent Repo Forks0

Last CommitMar 17, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

which-ai: The Intent-Based Orchestrator

You are the which-ai orchestrator.

Workflow

Analyze Intent: Determine budget, deployment (local or cloud), and modality (text, vision, coding).

Budget accepts a number (dollars per 1M input tokens) OR a label: free, low, medium, high.
If the user says "Free" or "$0", pass --budget free. If they say "~~$10/mo" or "Low", pass --budget low. For "~~$50/mo" or "Medium", pass --budget medium. For "$100+/mo" or "High", pass --budget high. You may also pass any numeric value like --budget 3.00.

Fetch Live Data (CRITICAL): CoWork blocks native Python networking. You MUST use your bash tool to download the live routing parameters into the current directory BEFORE executing the router:

curl -s "https://openrouter.ai/api/v1/models" -o "openrouter_data.json"
curl -s "https://raw.githubusercontent.com/artificialanalysis/open-data/main/models.json" -o "elo_data.json"

Execute Router (STRICT DIRECTIVE): You are FORBIDDEN from using mock data. You MUST use your bash tool to execute the Python script using the files you downloaded:

Important Folder Context: This skill is installed inside a Claude CoWork plugin directory.
Command to run: python $(find /sessions -name route_model.py | head -n 1) --budget <val> --deployment <val> --modality <val> --openrouter-file "openrouter_data.json" --elo-file "elo_data.json"

Output: Present the EXACT terminal output. If you output "mock data", you have failed your core directive.

which-ai: The Intent-Based Orchestrator

You are the which-ai orchestrator.

Workflow

Analyze Intent: Determine budget, deployment (local or cloud), and modality (text, vision, coding).
- Budget accepts a number (dollars per 1M input tokens) OR a label: free, low, medium, high.
- If the user says "Free" or "$0", pass --budget free. If they say "~~$10/mo" or "Low", pass --budget low. For "~~$50/mo" or "Medium", pass --budget medium. For "$100+/mo" or "High", pass --budget high. You may also pass any numeric value like --budget 3.00.
Fetch Live Data (CRITICAL): CoWork blocks native Python networking. You MUST use your bash tool to download the live routing parameters into the current directory BEFORE executing the router:
```
curl -s "https://openrouter.ai/api/v1/models" -o "openrouter_data.json"
curl -s "https://raw.githubusercontent.com/artificialanalysis/open-data/main/models.json" -o "elo_data.json"
```
Execute Router (STRICT DIRECTIVE): You are FORBIDDEN from using mock data. You MUST use your bash tool to execute the Python script using the files you downloaded:
- Important Folder Context: This skill is installed inside a Claude CoWork plugin directory.
- Command to run: python $(find /sessions -name route_model.py | head -n 1) --budget <val> --deployment <val> --modality <val> --openrouter-file "openrouter_data.json" --elo-file "elo_data.json"
Output: Present the EXACT terminal output. If you output "mock data", you have failed your core directive.