Production-ready prompt library for Xiaomi MiMo V2.5
120+ hand-tuned prompts across Chat, Reasoning, Image, and Speech models. Battle-tested for coding, Web3, security audits, content creation, and data analysis.
A static, client-side prompt library purpose-built for the Xiaomi MiMo V2.5 model family. Every prompt is hand-tuned, model-aware, and copy-paste ready for production use.
Built for developers integrating MiMo into apps, agents, and workflows via Claude Code, Cursor, Hermes Agent, or direct API calls.
The official MiMo docs cover the API surface. They don't cover how to prompt MiMo well — and each model in the V2.5 family responds best to fundamentally different patterns.
| Model | Optimized for | Reward |
|---|---|---|
| V2.5 Chat | Direct instruction, structure-first | Speed, low cost, high throughput |
| V2.5 Reasoning | Multi-step decomposition, chain-of-thought | Complex logic, RCA, deep analysis |
| V2.5 Image | Composition, lighting, lens vocabulary | Production-grade visuals |
| V2.5 Speech | Tone, pacing, emphasis cues | Natural TTS, accurate STT |
This library encodes those patterns so you ship faster instead of iterating from scratch.
- Open the live site
- Search by keyword (
Ctrl+K) or filter by model/category - Click a card to expand → see system + user + recommended params
- Click Copy → paste into your API call
- Replace
{{placeholders}}with your data
from openai import OpenAI
import os
client = OpenAI(
api_key=os.environ["MIMO_API_KEY"],
base_url="https://platform.xiaomimimo.com/v1",
)
response = client.chat.completions.create(
model="mimo-v2.5-reasoning",
messages=[
{"role": "system", "content": "You are a senior Web3 security auditor..."},
{"role": "user", "content": "Audit this Solidity contract: ..."},
],
temperature=0.2,
max_tokens=4000,
)import OpenAI from 'openai';
const client = new OpenAI({
apiKey: process.env.MIMO_API_KEY,
baseURL: 'https://platform.xiaomimimo.com/v1',
});
const completion = await client.chat.completions.create({
model: 'mimo-v2.5-chat',
messages: [
{ role: 'system', content: '...' },
{ role: 'user', content: '...' },
],
});120 prompts across 20 categories, organized by use case:
|
🛠️ Engineering
|
📊 Analysis & Reasoning
|
✍️ Content & Creative
|
💼 Productivity
|
V2.5 Reasoning ████████████████████████████ 63 prompts
V2.5 Chat ███████████████████ 42 prompts
V2.5 Image ████ 10 prompts
V2.5 Speech ██ 5 prompts
| Prompt | Category | Model |
|---|---|---|
| Solidity smart contract security audit | Coding | Reasoning |
| Bug root-cause analysis | Coding | Reasoning |
| Wallet approval audit | Web3 | Reasoning |
| ERC-20 token scam check | Web3 | Reasoning |
| Generate unit tests with edge cases | Coding | Reasoning |
| Twitter/X thread from a topic | Writing | Chat |
| Cluster user feedback into themes | Analysis | Reasoning |
| Optimize a slow SQL query | Coding | Reasoning |
| Polish a Farcaster cast | Farcaster | Chat |
| Cold outreach email | Chat |
- 🔍 Smart search — multi-word AND matching across title, tags, and prompt body
- ⌨️ Keyboard shortcuts —
Ctrl+Kto focus,Escto clear - 🏷️ Faceted filters — combine model + category + search
- ✨ Live highlight — search terms highlighted in real-time
- 📋 One-click copy — clipboard with visual confirmation
- 🌑 Dark, glassmorphic UI — optimized for desktop and mobile
- ⚡ Zero dependencies — pure HTML/CSS/JS, no build step
- 🔒 No signup, no telemetry, no cookies — just open and use
Pull requests welcome. Add your battle-tested prompts to help the community.
- Fork this repository
- Add an entry to
prompts.jsfollowing the schema:
{
id: "my-prompt-slug", // unique kebab-case
title: "My Prompt Title", // shown on the card
category: "Coding", // pick from existing or propose new
model: "V2.5 Reasoning", // V2.5 Chat | Reasoning | Image | Speech
useCase: "When to use this — one specific scenario.",
tags: ["python", "refactor"],
params: { temperature: 0.3, max_tokens: 2000 },
system: "Optional system message.",
user: "Prompt body. Use {{placeholder}} for variables.",
expectedOutput: "Brief description of what good output looks like."
}- Open a PR — we review and merge weekly
- ✅ Prompt tested at least once with MiMo V2.5 (include a screenshot/output sample in PR)
- ✅
useCaseis a specific real scenario — not generic ("do X with AI") - ✅
paramsreflect what worked, not defaults - ✅ Use
{{placeholders}}for user-injectable variables - ✅ Search existing prompts to avoid duplication
- ❌ No marketing fluff, no emoji-heavy outputs unless purposeful
- Multi-word smart search with highlight
- Mobile UI overhaul (2x2 stats, scroll chips, safe-area aware)
- Keyboard shortcuts (
Ctrl+K,Esc) - Per-prompt deep-link share URL (
?p=prompt-id) - Lunr.js full-text search index
- PR template + automated schema validation
- Indonesian / Chinese translations
- CLI:
npx mimoprompts <id>to copy directly to clipboard - VS Code extension: insert prompt at cursor
- Prompt versioning + changelog per prompt
- Pure static — HTML, CSS, vanilla JavaScript (no framework)
- Zero runtime dependencies — no npm install, no build step
- Hosted on GitHub Pages — free, fast, HTTPS, global CDN
- No telemetry, no cookies, no analytics — privacy by default
- WCAG-aware — semantic HTML, ARIA labels, keyboard navigation
mimoprompts/
├── index.html Entry point
├── styles.css Dark glassmorphic theme + responsive
├── app.js Search, filter, render, copy logic
├── prompts.js All 120 prompt entries
└── README.md You are here
- Xiaomi MiMo Platform — official API docs and console
- MiMo 100T Token Creator Program — incentive program
- Hermes Agent — local AI agent framework
- Claude Code — Anthropic's CLI coding agent
- Cursor — AI code editor
Not affiliated with Xiaomi. MiMo is a trademark of Xiaomi Inc.
Built for the Xiaomi MiMo Orbit 100T Token Creator Incentive Program · Made by @gyoomei
⭐ Star this repo if MimoPrompts saved you prompt-engineering time.