Practical overview
When it fits
Use Farfalle when you want a configurable web-search-and-answer application and need control over its search provider or language-model setup. Choose its local setup when you can run Docker and, for local answer generation, Ollama.
Who it is for
- People who want AI-generated answers grounded in searches across configurable web-search providers.
- Self-hosters who want to choose between Ollama models, cloud models or custom models through LiteLLM.
- Developers setting up an AI search interface or browser search entry for a local instance.
Documented uses
- Ask a question and have Farfalle plan and execute web searches before generating an answer.
- Choose a search backend such as SearXNG, Tavily, Serper or Bing.
- Generate answers with supported local Ollama models, cloud models or custom LiteLLM-connected models.
- Use a local Farfalle instance as a browser search engine.
Requirements & limits
The README lists a hosted demo at farfalle.dev and says it uses cloud models only. It also documents running the application locally with Docker Compose.
- The local development setup requires Docker.
- Using local answer models additionally requires Ollama and one of the listed supported models.
- Search-provider and cloud-model credentials depend on the chosen integrations; the README marks several keys optional and says they are not required when using Ollama.
- The documented frontend is Next.js and the backend is FastAPI; the repository lists TypeScript as its primary language and an Apache-2.0 license.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up Farfalle locally with Docker Compose
Prepare and start a local Farfalle instance, with Ollama available if using local answer models.
Before you start
- Docker
- For local models: Ollama, one supported model (Llama 3, Mistral, Gemma or Phi 3), and the Ollama server running
- Optional API keys for Tavily, Serper, OpenAI, Bing or Groq
Steps
Clone the Farfalle repository and enter its directory.
Copy
.env-templateto.envand add any API keys you plan to use; the README says keys are not required when using Ollama.Start the development containers with
docker-compose -f docker-compose.dev.yaml up -d.After startup, visit
http://localhost:3000.
Expected outcome
A locally hosted Farfalle web interface at the documented local URL.
Guide limitations
- Local model use requires Ollama to be running.
- The instructions describe the development Docker Compose setup.
Add a local Farfalle instance as a browser search engine
Point a browser search-engine entry at a locally running Farfalle instance.
Before you start
- A Farfalle instance running at
http://localhost:3000 - A browser with search-engine settings
Steps
Open the browser's settings and go to its Search Engines section.
Create a new search-engine entry using
http://localhost:3000/?q=%s.Add the entry as a search engine.
Expected outcome
The browser has a search-engine entry that sends the query to the local Farfalle URL.
Guide limitations
- The documented URL uses localhost, so it refers to an instance accessible on the same machine.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- rashadphz/farfalle · Official homepage title · retained in its source language