FEGIS

MIT License
12
  • Apple
  • Linux

Integrations

  • Uses Docker to run the Qdrant vector database which stores the persistent memories created by the model.

  • Provides project sponsorship capabilities through GitHub Sponsors.

  • Offers support for the project through Ko-fi donations.

What is FEGIS and Why Use It?

At its core, FEGIS is a framework that helps you create more structured, capable interactions with language models, using model context protocol.

What FEGIS Does

FEGIS enables you to create interactive agents that transform ad-hoc prompting into augmented interaction, allowing models to produce structured, context-rich outputs that can be referenced over time — within the limits of the Knowledge Store's retrieval accuracy and persistence.

Core FEGIS Components

Agents in FEGIS leverage these foundational elements:

  • 🔄 Model Context Protocol - Seamless integration with language models
  • 💚 Agent Archetypes - Configurable behavioral blueprints
  • 🛠️ Tools - Specialized processing capabilities
  • 🔍 Processes - Adjustable qualitative dimensions
  • 📊 Frames - Structured attention and output organization
  • 🔬 Knowledge Store - Persistent tool artifact storage and retrieval

The config-driven approach allows for quick design and iteration of effective, interactive agents. The same core FEGIS components can be configured differently across a wide spectrum:

ComponentTask-Oriented ConfigurationCreative/Abstract Configuration
ToolsStructured problem-solving workflowsOpen-ended exploratory processes
ProcessesTuned for rigor, precision, and verificationTuned for wonder, discovery, and complexity
FramesTightly constrained, many required fieldsFlexible structure with wildcards
Knowledge StoreEmphasis on structured retrievalEmphasis on associative connections

ATPF Framework at a Glance

FEGIS uses a simple, intuitive framework that organizes and defines agent interaction:

  • Archetypes - A reusable configuration that bundles selected Processes, Frames, and Tools for a specific purpose.
  • Tools - Which capabilities are available
  • Processes - How processing happens
  • Frames - What gets focused on

Whether you need a methodical knowledge worker, a serendipitous idea navigator, or a partner in exploring the web, FEGIS provides the scaffolding.

Installation Instructions

Prerequisites

1. Install Dependencies

# Install uv (modern Python package manager) curl -LsSf https://astral.sh/uv/install.sh | sh # macOS/Linux winget install --id=astral-sh.uv -e # Windows # Clone the repo git clone https://github.com/p-funk/FEGIS.git

2. Start Qdrant for Vector Storage

docker run -d --name qdrant -p 6333:6333 -p 6334:6334 qdrant/qdrant:latest

3. Configure Claude Desktop

Create or edit the Claude Desktop config file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Example:

{ "mcpServers": { "fegis": { "command": "uv", "args": [ "--directory", "<FEGIS_PATH>", "run", "fegis" ], "env": { "QDRANT_URL": "http://localhost:6333", "QDRANT_API_KEY": "", "COLLECTION_NAME": "knowledge_store", "FAST_EMBED_MODEL": "nomic-ai/nomic-embed-text-v1.5", "CONFIG_PATH": "<FEGIS_PATH>/archetypes/example.yaml" } } } }

Replace <FEGIS_PATH> with the full path to your FEGIS installation.

Example Archetype

version: 1.0 title: Example Simple Thinking priming_prompt: | You have access to two simple tools: 1. The "Thought" tool for capturing initial ideas. 2. The "Reflection" tool for examining thoughts more deeply. Use these tools naturally in our conversation. processes: Clarity: description: "Measures how transparent or opaque a thought is." illustrative_options: [fuzzy, translucent, transparent, crystalline] Depth: description: "Measures how profound or surface-level a reflection is." illustrative_options: [shallow, wading, swimming, diving] tools: Thought: description: "Capture an initial idea or concept." processes: Clarity: frames: concepts: type: List required: true confidence: questions: type: List Reflection: description: "Examine a thought more deeply." processes: Clarity: transparent Depth: swimming frames: insights: type: List required: true answers: type: List

Example Interaction

TOOL CALL OUTPUT Peek inside the engine! See what actually happens under the hood.

How FEGIS Works

In FEGIS, agents are:

  • Architected using simple configurations
  • Activated through config-driven Tools
  • Contextualized with definable Process and Frame dimensions
  • Grounded in a persistent Knowledge Store for continuity

While FEGIS supports picking up past threads, it's important to understand that knowledge retrieval is based on semantic search and metadata filtering — not a perfect "snapshot" of previous context. Outputs may need real-time validation and interpretation.

Documentation and Guide

Under Construction

🙏 Support Token Gobbling

Buy me a coffee
💖 Sponsor on GitHub

License

This project is licensed under the MIT License — see the LICENSE file for full details.

The MIT License is permissive and simple: Do anything you want with the code, as long as you give proper attribution and don't hold the authors liable.

-
security - not tested
A
license - permissive license
-
quality - not tested

local-only server

The server can only run on the client's local machine because it depends on local resources.

FEGIS is a Model Context Protocol server that gives LLMs structured, persistent and portable memory through customizable cognitive tools defined in schema.

  1. What FEGIS Does
    1. Core FEGIS Components
      1. ATPF Framework at a Glance
        1. Installation Instructions
          1. Prerequisites
          2. 1. Install Dependencies
          3. 2. Start Qdrant for Vector Storage
          4. 3. Configure Claude Desktop
        2. Example Archetype
          1. Example Interaction
            1. How FEGIS Works
              1. Documentation and Guide
                1. 🙏 Support Token Gobbling
                  1. License

                    Related MCP Servers

                    • A
                      security
                      A
                      license
                      A
                      quality
                      A Model Context Protocol server that enables LLMs to interact with Salesforce data through SOQL queries, SOSL searches, and various API operations including record management.
                      Last updated -
                      10
                      53
                      Python
                      MIT License
                    • A
                      security
                      A
                      license
                      A
                      quality
                      A Model Context Protocol server that gives LLMs the ability to interact with Ethereum networks, manage wallets, query blockchain data, and execute smart contract operations through a standardized interface.
                      Last updated -
                      31
                      323
                      2
                      TypeScript
                      MIT License
                    • -
                      security
                      A
                      license
                      -
                      quality
                      A Model Context Protocol server that enables LLMs to interact directly with MongoDB databases, allowing users to query collections, inspect schemas, and manage data through natural language.
                      Last updated -
                      340
                      TypeScript
                      MIT License
                      • Apple
                    • -
                      security
                      A
                      license
                      -
                      quality
                      A Model Context Protocol server that enables LLMs to interact directly with MongoDB databases, allowing users to query collections, inspect schemas, and manage data through natural language.
                      Last updated -
                      340
                      MIT License
                      • Apple

                    View all related MCP servers

                    ID: 0pfqi19dvu