A quick tip: Claude can automate Blender to convert 2D images into 3D models.
The entire process flows seamlessly.
Moreover, with just a single prompt, you can build an interactive webpage based on that scene.
The key behind this is the recently popular MCP (Model Context Protocol)—the crucial trick used to replicate Manus’s capabilities.
By integrating this protocol with Blender, you can achieve the results shown above.
What once took humans hours of manual modeling work can now be completed in minutes without any human intervention.
The open-source project BlenderMCP has already garnered 3.8k stars on GitHub within just three days of its release.
Furthermore, the modeling results are reliable. One user tested it by asking Claude to design Martian terrain; Claude was able to handle errors and issues independently while keeping humans informed.
No wonder some observers exclaimed: “Humans may no longer need design tools anymore—amazing!”
AI combined with application tools is becoming increasingly powerful.
Notably, this approach can be replicated across other open-source professional tools.
For instance, someone has already implemented MCP + QGIS (the geographic equivalent of “Photoshop”) to automate sensor mapping using Claude.
“Blender and Cursor Can Both Use MCP”
In simple terms, BlenderMCP connects Blender to Claude, allowing Claude to directly interact with and control Blender.
Based on BlenderMCP, many other tasks can be accomplished.
For example, creating a dungeon scene guarded by a dragon protecting a pot of gold.
Prompt: Create a low poly scene in a dungeon, with a dragon guarding a pot of gold.
During this process, instruction following performed well.
The prompt specifically emphasized “low poly,” and the final result indeed featured roundish dragons and pots.
It can also be used to build realistic beach scenes.
Prompt: Create a beach vibe using HDRIs, textures, and models like rocks and vegetation from Poly Haven.
This instruction requires modeling a beach using HDRIs, textures, and assets such as rocks and vegetation sourced from Poly Haven.
Poly Haven is a free, open-source 3D resource website. As seen in the video, Claude can directly download and utilize these resources on its own.
Other capabilities that can be tested include:
- “Paint this car red with a metallic finish.”
- “Create a sphere and place it above a cube.”
- “Set the lighting to a studio effect.”
- “Aim the camera at the scene and set it to an isometric view.”
According to the project page, BlenderMCP enables capabilities such as creating, modifying, and deleting 3D objects; using and modifying materials and colors; scene inspection; and code execution.
The system primarily consists of two parts: a Blender Addon and an MCP Server.
The former is a Blender plugin that creates a server within Blender to accept and execute commands. The latter implements the MCP protocol.
Detailed installation instructions have been fully open-sourced on GitHub by the author.
Beyond integrating MCP into Blender, netizens are also experimenting with using it to upgrade various other tools.
Even AI coding software becomes more automated when utilizing MCP.
One user employed the MCP protocol on Cursor to simultaneously integrate Slack and GitHub, completing a new feature development cycle.
After configuring the plugin and completing authentication, Cursor automatically read requirement documents from Slack via MCP, pulled code from GitHub, and autonomously wrote and uploaded the new feature.
This workflow leverages an MCP service provided by Composio, which can be configured directly within Cursor via a link.
Composio has also turned GitHub, Google Search, email, maps, and more into MCP services.
Besides Composio, MCP enthusiasts have established their own communities, providing a vast array of open-source server and client resources.
For example, this particular MCP service allows searching for papers on arXiv; after following the tutorial to configure it, users can directly search for papers within the Claude interface.
Interestingly, large language models themselves can be “MCP-enabled.” For instance, a server can call other models via an OpenAI-compatible API.
It is even possible to integrate DeepSeek-R1 into Claude without issue.
Why is MCP Truly Powerful?
MCP is a communication protocol proposed by Anthropic. Anthropic currently likens it to the Type-C interface for AI applications.
Furthermore, Anthropic intends to lead efforts to establish MCP as an open industry standard.
It enables seamless integration between large language model applications and external data sources and tools, helping AI acquire the necessary contextual data to generate higher-quality, more task-relevant responses.
MCP primarily addresses a common pain point facing application developers globally: data isolation.
It acts as a bridge between AI systems and data sources, allowing developers to establish bidirectional connections between data repositories and AI tools.
MCP adopts a client-server architecture, where multiple services can connect to any compatible client. Clients may include Claude Desktop, IDEs, or other AI tools, while servers act as adapters that expose data sources.
Its key advantage is that whether accessing local resources (databases, files, services) or remote resources (such as Slack or GitHub APIs), the same protocol can be used.
Moreover, it supports a wide variety of data formats, including file content, database records, API responses, real-time system data, screenshots and images, log files, etc., covering almost all types.
MCP servers also incorporate built-in security mechanisms, allowing the server to control resources directly without requiring developers to hand over API keys to the large language model.
Depending on the service source, MCP primarily uses communication mechanisms: standard input/output for local communication and Server-Sent Events (SSE) for remote communication.
Messages in both communication modes utilize JSON format for transmission, enabling standardized MCP communication and providing scalability.
Although the services callable by MCP appear numerous and complex, the development process is actually quite simple.
In its official announcement at launch, it was explicitly stated that the then-latest Claude 3.5 Sonnet was already highly proficient in setting up MCP servers, effectively completing the loop directly.
With powerful calling capabilities, a convenient development workflow, backing from Anthropic, and attention from the open-source community, MCP seems poised to become a future AI standard as envisioned by Anthropic.
But can it truly achieve this?
There are indeed quite a few people holding a wait-and-see or pessimistic attitude.
Recently, LangChain, a well-known open-source large model framework, conducted a poll on X (formerly Twitter).
40.8% of respondents believe MCP will become the future standard, while more people feel it is best to wait and see.
Disagreements have also emerged within LangChain itself.
The CEO believes that MCP lowers the barrier for agents to integrate with tools.
However, a founding engineer argues that from an engineering perspective, many customized requirements will arise, and in many cases, MCP cannot fully deliver its potential.
For MCP to live up to its hype, it needs to become as popular as OpenAI’s GPTs; however, GPTs themselves do not seem to be particularly well-received.
What do you think? Will MCP be a flash in the pan?
Feel free to leave comments and discuss below~
GitHub repository:
https://github.com/ahujasid/blender-mcp?tab=readme-ov-file
References
- 1900240156826939560 — x.com/bilawalsidhu/status/1900240156826939560
- 1900632591516008599 — x.com/bilawalsidhu/status/1900632591516008599
- 1898789901824590328 — x.com/mattpocockuk/status/1898789901824590328
- 1898439847322525963 — x.com/KaranVaidya6/status/1898439847322525963
- mcp fad or fixture — blog.langchain.dev/mcp-fad-or-fixture/