FUTURESCALE

Futurescale Founder Joins Model Context Protocol Steering Committee

An exciting new way to extend tool use to any large language model

The Model Context Protocol was created by Anthropic, the makers of Claude, and is maintained by their core team and a cadre of industry experts as a fully Open Source project. It offers:

• Industry-wide adoption
• Best practices for allowing LLMs to interact with your sensitive data
• Agnostic approach to LLM vendors, allowing you to use the right model for each task
• A fast-growing list of servers that integrate LLMs with your favorite cloud services
• SDK support in multiple languages, (Python, TypeScript, Java, Kotlin, C#, and more on the way)

Why MCP

Model Context Protocol Architecture

Frontier models like GPT-x, Claude, Gemini, etc. are continually one-upping each other with new capabilities and personalities. One day it seems as if OpenAI is at the head of the pack, then out of the blue, it's Deepseek. Placing all your chips on one platform is a risky bet for builders who are building intelligent agents and incorporating these models into their workflows.

MCP provides a way for all models, even ones small enough to run on your run-of-the-mill laptop, with the ability to connect to any API and take actions on your behalf. Logging into Slack and checking your messages, creating a pull request on Github, or collaborating with other agents on a larger task.

Why Futurescale

As previously reported we began building on the MCP almost immediately following its release. Our GooseTeam and Puzzlebox MCP servers are focused on agent collaboration and coordination with an eye toward long horizon tasks.

Block, makers of the popular agent Codename Goose wrote about GooseTeam on its blog, getting a lot of eyes on the project.

Later, when Puzzlebox required MCP features not yet testable with the official MCP Inspector, we added the them, and upgraded their example "everything" server with that support as well. This allowed developers to experiment with dynamic resource subscriptions, an important but still underutilized feature of the protocol.

After adding several useful features and fixes to the repos, we were invited to the MCP maintainers group and steering committee. Since then, responding to issues, reviewing and merging pull requests, adding features, fixing bugs, and participating in working group discussions has become nearly a full time job.

Nevertheless, progress proceeds on our own efforts, including the new desktop application Puzzlemaster, which will house teams of agents coordinated with Puzzlebox and collaborating with GooseTeam to tackle long horizon tasks.

We are honored to be among the industry experts leading the charge to make MCP an industry standard.