FRA — Feather Research Agent

FRA

Feather Research Agent — Private AI Research System

FRA is a self-hosted AI workspace you run on your own computer. It is not a cloud service, not a subscription, and does not compete with ChatGPT or Claude. You choose the AI model. You own your data. One command to install.

Get FRA — One Command

Requirements: Python 3.11+, Windows, Mac, or Linux.

# Clone and install

git clone https://github.com/azzizzybud/fra-sesh.git

cd odysseus

py -3.11 -m venv venv

venv\Scripts\Activate.ps1

pip install -r requirements.txt

python setup.py

# Start the server

python -m uvicorn app:app –host 127.0.0.1 –port 7000

Open http://localhost:7000 in your browser. Log in with the temporary password printed in the terminal. Go to Settings and add your AI models.

Connect Your Models

FRA Sesh works with any of these — add them in Settings:

  • DeepSeek — via OpenRouter
  • Claude — via Anthropic API
  • OpenAI — GPT-4o, o1, o3, etc.
  • Ollama — any local model (Llama, Mistral, Gemma)
  • Anything with an OpenAI-compatible API endpoint

What You Get

Chat Converse with any model you connect — DeepSeek, Claude, Ollama, anything
Agent Give it tools — files, shell, web, MCP — and it runs tasks autonomously
Deep Research Multi-step research that gathers sources and produces visual reports
Documents Multi-tab editor — markdown, HTML, CSV — with AI-assisted editing
Email + Calendar IMAP/SMTP inbox with AI triage. CalDAV sync with Apple, Fastmail, Nextcloud
Memory Persistent knowledge base — the system remembers across sessions
Mobile Responsive design works on your phone — installable as PWA

What’s Different

FRA Sesh is the interface console of the FRA system. On its own, it’s a powerful private AI workspace — you get chat, agents, research, documents, email, and calendar in one place, with any model you choose.

Combined with FRA Rekh (the autonomous research agent), it becomes a system that runs research in the background, governed by mathematical theorems — stability detection, evidence tracking, structural audit. But you don’t need any of that to start. The console works on its own.

Open Source

FRA Sesh is built on Odysseus (MIT licensed). Forked and branded as FRA Sesh. Full credit to the Odysseus project. The FRA system extends it with Feather governance — mathematical theorems as operating modules.

MIT License. Your data stays on your machine. No cloud, no sign-up, no subscription.

Ma’at is the measure.

Scroll to Top