❓What Is Fabelis?

🌟 Origins

Fabelis began as an AI agent framework built on rig.rs to extend its functionality into social media agents, story telling, narration, and a new platform for open-source collaboration in the budding AI space fostered by arc.

🚀 Evolution

Over time, the narrative around AI agents expanded from personality agents to arming agents with the tools to accomplish meaningful tasks. Fabelis was uniquely positioned as a framework to adopt the model context protocol (MCP for short), which effectively exposes an API layer to agents allowing them to accomplish meaningful tasks.

🛠️ The Fabelis Framework

Fabelis expresses itself mainly though its mcp-core and agent repositories. The former being the representation of our MCP client/server model and the extension of rig's core functionality to include this new powerful protocol. The latter being an opinionated extension and adaptation of Rig to allow for feature expansion.

We heavily encourage and incentivize community updates, feel free to send in a pull request if you have a good update or idea!

🔧What is MCP & How It Works

MCP can roughly be described as the API layer for agents. The protocol directly addresses the problem agents initially faced in that they were purely generative & limited in their operation by producing a layer of tools that agents can use when they deem necessary. This allows you to go from prompt to workflow. Imagine asking your Fabelis AI character to edit products in your shopify store based on its intuition In our repos we produce a few things: - MCP-Core repository & crate: Built in rust and fully compatible with any agent framework - a premier interface for Agent -> Workflow communications - MCP-RIg repository: Where Fabelis uniquely positions itself in the complex building on top of RIG to allow our favorite framework's agents to go out and do things. - Fabelis MCP Servers: Proprietary, but open to use servers that bring Fabelis agent functionality to the real world, also compatible with any agent framework or client regardless of programming language

We've thought long and hard about making agents useful and more than