Introduction
OpenCopilot is a free and open-source tool that allows you to create AI copilots for your SaaS product. The copilot can communicate with your underlying APIs and execute calls as needed. We are open-source under MIT license and also support OpenAPI standards.
Getting started
What are copilots? (start from here)
Introduction to OpenCopilot. start from here for step-by-step guide
Setting up
Install the project locally and make it yours
Create copilots
Create copilots programmatically (APIs, SDKs) or using the dashboard
Authentications and Authorization
Handle Authentications and Authorization to your backend
Embedding copilots on your app
How to use OpenCopilot widget and embed it on your web app
Creating complex flows
Learn how to use OpenCopilot flows definition
Roadmap
See what’s coming next.
Setting up
You can self-host OpenCopilot in a relatively easy way, please make sure you have the following requirements:
- Docker engine and docker compose installed
- Clone the repository
- Update your llm-server/.env with your OPENAI_API_KEY
- Run the installation script
Then your OpenCopilot dashboard will be accessible at http://localhost:8888
You also can see the complete list of commands using make help
Crawling Web Pages
By default, when passing in a URL to create a new DataSource for a Copilot, 15 pages will be indexed. The configuration for this is located in the opencopilot/llm-server/utils/llm_consts.py
file:
To increase the max pages to crawl, you can change the configuration file. However, this will create uncommitted changes in your working copy which you will need to stash in order to git pull
for future updates. Another way to increase the max pages to crawl is to set an environment variable:
Complete list of links:
- http://localhost:8888/backend -> backend APIs (create, update, delete and validate copilots)
- http://localhost:8888 or http://localhost:8888/dashbaord -> dashboard (UI to do the same)