Refactor high-complexity React components in Dify frontend. Use when `pnpm analyze-component...
npx skills add terry-li-hm/skills --skill "ask-llms"
Install specific skill from multi-skill repository
# Description
Query multiple LLMs in parallel via OpenRouter and compare their responses. Use when the user wants diverse perspectives from different models on a question.
# SKILL.md
name: ask-llms
description: Query multiple LLMs in parallel via OpenRouter and compare their responses. Use when the user wants diverse perspectives from different models on a question.
Ask LLMs
Query multiple LLMs simultaneously via OpenRouter and display their responses side-by-side. For quick parallel comparisons — use /llm-council when you want models to actually debate and synthesize.
When to Use
Use this skill when:
- Comparing how different LLMs answer the same question
- Need diverse perspectives on a problem or decision
- Want quick "second opinions" from multiple models
- Curious about model differences on a topic
- Quick feedback on outreach messages (see /outreach for style tips)
Prerequisites
OpenRouter API key (required):
export OPENROUTER_API_KEY=sk-or-v1-...
Instructions
Step 1: Get the Question
Ask the user what question they want answered, or use the question they already provided.
Step 2: Run the Query
Expensive mode (default) — frontier models with thinking:
cd /Users/terry/skills/ask-llms
uv run council.py "YOUR QUESTION HERE"
Cheap mode — fast and affordable:
uv run council.py "YOUR QUESTION HERE" --cheap
Step 3: Present Results
The script outputs each model's response with clear separators. Highlight:
- Key differences between models
- Points of agreement
- Which response is most relevant to the user's situation
Model Tiers
Expensive (default) — 2026 frontier models with thinking/reasoning:
- Claude Opus 4.5
- GPT-5.2
- Gemini 3 Pro
- Grok 4
- DeepSeek R1
Cheap (--cheap flag) — fast and affordable:
- Claude Sonnet 4.5
- GPT-4o
- Gemini Flash
- Grok 4.1 Fast
- DeepSeek V3.2
Options
| Flag | Description |
|---|---|
--cheap |
Use cheaper/faster models |
--no-thinking |
Disable reasoning mode for expensive models |
--timeout N |
Timeout in seconds (default: 60 cheap, 180 expensive) |
--models "a,b,c" |
Override with custom models |
vs /llm-council
| Aspect | /ask-llms | /llm-council |
|---|---|---|
| Pattern | Parallel queries | Deliberation + synthesis |
| Output | Side-by-side responses | Consensus with reasoning |
| Speed | Fast (~10-30s) | Slower (~60-90s) |
| Use case | Quick comparison | Important decisions |
Files
- Script:
/Users/terry/skills/ask-llms/council.py - This skill:
/Users/terry/skills/ask-llms/SKILL.md
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.