- unwind ai
- Posts
- Bundle Multiple MCP Servers into One
Bundle Multiple MCP Servers into One
PLUS: All-in-one Context Engineer in your IDE, Gemini with Deep Think earns gold at IMO
Shubham Saboo & Gargi Gupta
July 22, 2025
In partnership with
Today’s top AI Highlights:
Bundle multiple MCP servers into one unified MCP server
The first “Context Engineer” built into your IDE
Google Gemini with Deep Think “officially” earns gold at the IMO
Qwen’s model beats Claude 4 and Kimi K2 with just 25% parameters
AI agents to use your Android and iOS apps autonomously
& so much more!
Read time: 3 mins
AI Tutorial
Business consulting has always required deep market knowledge, strategic thinking, and the ability to synthesize complex information into actionable recommendations. Today's fast-paced business environment demands even more - real-time insights, data-driven strategies, and rapid response to market changes.
In this tutorial, we'll create a powerful AI business consultant using Google's Agent Development Kit (ADK) combined with Perplexity AI for real-time web research. This consultant will conduct market analysis, assess risks, and generate strategic recommendations backed by current data, all through a clean, interactive web interface.
We share hands-on tutorials like this every week, designed to help you stay ahead in the world of AI. If you're serious about leveling up your AI skills and staying ahead of the curve, subscribe now and be the first to access our latest tutorials.
Latest Developments
MCP servers are individually brilliant but collectively chaotic when you're trying to manage more than a handful.
You end up with scattered endpoints, inconsistent authentication, and the constant headache of figuring out which tools are actually being used by your AI applications.
MetaMCP solves this orchestration nightmare by acting as a unified aggregator that groups your MCP servers into organized namespaces, applies smart middleware, and serves everything through standardized endpoints. This opensource tool transforms your fragmented MCP infrastructure into a cohesive system with proper authentication, monitoring, and debugging capabilities.
Instead of juggling multiple server configurations and connections, you get a single Docker-powered solution that handles the complexity while you focus on building better AI applications.
Key Highlights:
Namespace - Group MCP servers into organized namespaces and expose them through SSE or streamable HTTP endpoints, making your MCP infrastructure actually manageable instead of a tangled mess.
Cherry-pick Tools - Pick only the tools you need when remixing MCP servers. Apply pluggable middleware for observability and security.
Cold Start Killer - Pre-warms idle sessions to eliminate those annoying cold start delays, plus includes a built-in MCP inspector for debugging without the usual configuration headaches.
Universal Compatibility - Works seamlessly with Claude Desktop, Cursor, Open WebUI, and any MCP client through various transport methods, complete with API key authentication and multi-tenancy support.
Create How-to Videos in Seconds with AI
Stop wasting time on repetitive explanations. Guidde’s AI creates stunning video guides in seconds—11x faster.
Turn boring docs into visual masterpieces
Save hours with AI-powered automation
Share or embed your guide anywhere
How it works: Click capture on the browser extension, and Guidde auto-generates step-by-step video guides with visuals, voiceover, and a call to action.
The conversation has shifted from prompt engineering to context engineering - a broader, more powerful concept where success depends on providing the right information and tools, in the right format, at the right time.
It's about creating predictable, repeatable interactions that maximize AI's ability to understand and respond appropriately.
ByteDance just released TRAE SOLO, an all-in-one “Context Engineer” integrated directly into your IDE, making it the first major product built entirely around this paradigm shift rather than prompt crafting. SOLO doesn't just respond to your requests, it actively gathers the right context from your project architecture, documentation, and real-time system outputs to think, plan, build, and ship complete software features autonomously.
Key Highlights:
Autonomous SOLO Mode - Handles your entire development workflow from requirements to deployment while you monitor progress through an integrated view of editor, browser, terminal, and documentation. The agent works independently, writing code, fixing bugs, and shipping features without constant human intervention.
@SOLO Builder for Full-Stack - Built-in web developer that transforms concepts into fully functional websites with end-to-end implementation, handling everything from PRD drafting to UI design, code generation, and deployment for both professional developers and hobbyist builders.
Mode Switching - Granular control lets you choose between letting the agent work completely autonomously in SOLO mode or keeping it in the chat panel with traditional IDE mode. One simple switch on the top left gives you full control over how much the AI takes over.
Context Engineering at Scale - Uses voice input, gathers real-time context from system outputs, and maintains memory across the extended view to provide complete project understanding rather than relying on single prompts like traditional AI coding tools.
TRAE SOLO is included in the Pro Plan but users need a SOLO Code to activate SOLO Mode - watch for official drops on Trae's social channels and partner events for code giveaways.
Quick Bites
Alibaba Qwen cooked hard this time! They ditched their Qwen3-235B-A22B with hybrid thinking, and released an updated model instruct model. With just 22B active parameters, this model outperforms Claude 4 Opus and the recent opensource king Kimi K2 across tasks like reasoning, math, and coding. The model is already live on Qwen Chat and available in both standard and FP8 quantized versions for immediate deployment.
The math Olympics just got its second AI gold medalist in 24 hours. Google DeepMind confirmed an advanced version of their Gemini with Deep Think also achieved gold medal performance on the 2025 IMO, solving 5 out of 6 problems for 35 points, exactly matching OpenAI's score but with official IMO coordinator grading and certification. Unlike last year's AlphaProof system that required expert translation into formal languages and days of computation, this version operated end-to-end in natural language within the 4.5-hour competition time limit. This Deep Think model will be available to a small group of testers before Google AI Ultra subscribers.
Tools of the Trade
DroidRun: Allow AI agents to autonomously use Android and iOS phones with just natural language prompts. Uses screenshot analysis and vision models to understand device UI. Supports multiple LLM providers (OpenAI, Anthropic, Gemini).
Async: Coding agents that handle gruntwork like bug fixes, migrations, and maintenance work that teams don't want to do manually. The agents run asynchronously in the cloud in parallel, and submit pull requests for review after completing tasks. Delegate and review from your phone.
Claude Memory Visualizer: Converts Claude's memory.json files into interactive network graphs with zero setup. You can explore entities, relationships, and observations through a physics-based visualization that automatically finds your memory file.
Awesome LLM Apps: Build awesome LLM apps with RAG, AI agents, MCP, and more to interact with data sources like GitHub, Gmail, PDFs, and YouTube videos, and automate complex work.
Hot Takes
unpopular take:
virality >> talking to users
10,000 users’ worth of data is a lot more helpful than talking to 3 users
it’s just not recommended because not everyone can get 10,000 users’ worth of data immediately
except thx to short form, u can now
think more ab going viral ~
Roy LeeOld inequality: "I have access to knowledge, you don't."
New inequality: "We both have infinite knowledge. I know what to do with it." ~
Emad Mostaque
That’s all for today! See you tomorrow with more such AI-filled content.
Don’t forget to share this newsletter on your social channels and tag Unwind AI to support us!
PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one, two (or 20) of your friends 😉
Reply