My story, I’m a cloud security architect and was evaluating AWS’ Bedrock Agent Core runtime for agents. I built a very basic AI agent using strands and all was working fine using Bedrock as the model provider until an invisible token limit kicked in. Further more, I found the end to end ceremony of using the Bedrock Agent Core Starter toolkit, setting up permissions in AWS, such an ordeal and thought there must be an easier way to get up and running with Agents. So I build the Agentify initially as a library of wrapper functions to LLM model providers and then evolved it into a CLI and began building components so you can run, serve (as API HTTP endpoint)or deploy to a local agent runtime. Using YAML files to declare both agents and tools got me up and running quickly. So the project is cloud agnostic, framework agnostic and model agnostic, it is about it learning, experimenting and prototyping. I believe strongly that in order to secure something, you need to know how it works, how it breaks, and how it can be misused. What I’ve built is not another framework or orchestration toolkit, it’s for local testing. Appreciate any feedback for how I can improve it. Thanks, Lewis
My story, I’m a cloud security architect and was evaluating AWS’ Bedrock Agent Core runtime for agents. I built a very basic AI agent using strands and all was working fine using Bedrock as the model provider until an invisible token limit kicked in. Further more, I found the end to end ceremony of using the Bedrock Agent Core Starter toolkit, setting up permissions in AWS, such an ordeal and thought there must be an easier way to get up and running with Agents. So I build the Agentify initially as a library of wrapper functions to LLM model providers and then evolved it into a CLI and began building components so you can run, serve (as API HTTP endpoint)or deploy to a local agent runtime. Using YAML files to declare both agents and tools got me up and running quickly. So the project is cloud agnostic, framework agnostic and model agnostic, it is about it learning, experimenting and prototyping. I believe strongly that in order to secure something, you need to know how it works, how it breaks, and how it can be misused. What I’ve built is not another framework or orchestration toolkit, it’s for local testing. Appreciate any feedback for how I can improve it. Thanks, Lewis