• Core Architecture and Licensing Clarity
• Getting started
  • Local (Native) Mode
    • 1 · Prerequisites
    • 2 · Clone
    • 3 · Add your OpenAI key
    • 4 · Run the services
    • Advanced developer tips
      • Prerequisites
  • Dev-Container mode
• Advanced configuration
  • Supported Model Providers
  • Configuration Files
  • System Architecture
  • Advanced Integrations
• Documentation
• Contributing
• License
• Contacts

Fred is both:

• An innovation lab — to help developers rapidly explore agentic patterns, domain-specific logic, and custom tools.
• A production-ready platform — already integrated with real enterprise constraints: auth, security, document lifecycle, and deployment best practices.

It is composed of:

• a Python agentic backend (FastAPI + LangGraph)
• a Python knowledge flow backend (FastAPI) for document ingestion and vector search
• a React frontend

Fred is not a framework, but a full reference implementation that shows how to build practical multi-agent applications with LangChain and LangGraph. Agents cooperate to answer technical, context-aware questions.

See the project site: https://fredk8.dev

The three components just described form the entirety of the Fred platform. They are self-contained and do not require any external dependencies such as MinIO, OpenSearch, or Weaviate.

Instead, Fred is designed with a modular architecture that allows optional integration with these technologies. By default, a minimal Fred deployment can use just the local filesystem for all storage needs.

Fred is released under the Apache License 2.0. It does *not embed or depend on any LGPLv3 or copyleft-licensed components. Optional integrations (like OpenSearch or Weaviate) are configured externally and do not contaminate Fred's licensing. This ensures maximum freedom and clarity for commercial and internal use.

In short: Fred is 100% Apache 2.0, and you stay in full control of any additional components.

Fred works out of the box when you provide one secret — your OpenAI API key.
Defaults:

• Keycloak is bypassed by a mock admin/admin user
• All data (metrics, conversations, uploads) is stored on the local filesystem
• No external services are required

Production services and databases can be added later or via the deployment factory repository.

git clone https://github.com/ThalesGroup/fred.git
cd fred

echo "OPENAI_API_KEY=sk-..." > {agentic_backend,knowledge_flow_backend}/config/.env

# Terminal 1 – agentic backend
cd agentic_backend && make run

# Terminal 2 – knowledge flow backend
cd knowledge_flow_backend && make run

# Terminal 3 – frontend
cd frontend && make run

Open http://localhost:5173 in your browser.

To get full VS Code Python support (linting, IntelliSense, debugging, etc.) across our repo, we provide:

1. A VS Code workspace file fred.code-workspace that loads all sub‑projects.
2. Per‑folder .vscode/settings.json files in each Python backend to pin the interpreter.

• Visual Studio Code
• VS Code extensions:
  • Python (ms-python.python)
  • Pylance (ms-python.vscode-pylance)

1. Open the workspace

After cloning the repo, you can open Fred's VS Code workspace with code fred.code-workspace

When you open Fred's VS Code workspace, VS Code will load four folders:

• fred – for any repo‑wide scripts
• agentic_backend – first Python backend
• knowledge_flow_backend – second Python backend
• frontend – UI

2. Per‑folder Python interpreters Each backend ships its own virtual environment under .venv. We’ve added a per‑folder VS Code setting (see for instance agentic_backend/.vscode/settings.json) to automatically pick it:

This ensures that as soon as you open a Python file under agentic_backend/ (or knowledge_flow_backend/), VS Code will:

• Activate that folder’s virtual environment
• Provide linting, IntelliSense, formatting, and debugging using the correct Python

If you prefer a fully containerised IDE with all dependencies running:

1. Install Docker, VS Code (or an equivalent IDE that supports Dev Containers), and the Dev Containers extension.
2. Create ~/.fred/openai-api-key.env containing OPENAI_API_KEY=sk-….
3. In VS Code, press F1 → Dev Containers: Reopen in Container.

The Dev Container starts the devcontainer service plus Postgres, OpenSearch, and MinIO. Ports 8000 (backend) and 5173 (frontend) are forwarded automatically.

Inside the container, start the servers:

# Terminal 1 – agentic backend
cd agentic_backend && make run

# Terminal 2 – knowledge flow backend
cd knowledge_flow_backend && make run

# Terminal 3 – frontend
cd frontend && make run

See agentic_backend/config/configuration.yaml (section ai:) for concrete examples.

• Enable Keycloak or another OIDC provider for authentication
• Persist metrics and files in OpenSearch and MinIO

• Main docs: https://fredk8.dev/docs
• Agentic backend README
• Frontend README
• Knowledge Flow backend README
• Code of Conduct
• License
• Security
• Python Coding Guide
• Contributing

We welcome pull requests and issues. Start with the Contributing guide.

Apache 2.0 — see LICENSE

• alban.capitant@thalesgroup.com - fabien.le-solliec@thalesgroup.com - florian.muller@thalesgroup.com - simon.cariou@thalesgroup.com  - dimitri.tombroff@thalesgroup.com