Uploading your own agent framework

Guide

1. Click on “Upload New Framework”.

1. Complete the following steps and review the summary panel before clicking “Submit”

   1. Enter your Agent Framework name (lowercase letters only, e.g. helloworld)

   2. Write a description of the AI Agent Framework (e.g., “ A sophisticated furry chat agent specializing in handling mission-critical conversations”)

   3. Input the Docker Image(e.g., docker_org_name/imagine:tag)

   4. Input Agent Logo (e.g. https://example.com/ottie007.png)

   5. Choose your Chat Endpoint (e.g. /poke)

   6. Input Port ( e.g. 3000)

   7. Ensure your agent chat framework follows the Chat Request and Chat Response Schema as shown below.

      
   8. Input GitHub URL (e.g. https://github.com/alt-research)

   9. Specify Environment Variables (optional)

   10. Finally, ensure all details are correct in the Summary panel and click on “Submit”.

2. You can now find your uploaded framework under "My Uploads" tab. If you wish to, you can opt to "Make Public" your framework in our Autonome App Store by submitting your request. Typically, these requests will be approved in two to three business days if there are no issues.

   

Troubleshooting bad deployments with uploaded frameworks

If your agent deployment doesn't work with your uploaded framework, please do ensure your framework meets the following criteria before uploading again:

• It has to be built for Linux architecture, you can use docker buildx build --platform=linux/amd64 if you are on MacOS;

• The docker image url should be publicly accessible in order for our infrastructure to pull the image (https://hub.docker.com/);

• The agent must have some API endpoints available to communicate, instead of being just a CLI application