Prime Intellect Releases Prime Agent: An Open-Source RLM Harness Where Sub-Agents Are Function Calls Inside Persistent IPython Kernel

Prime Intellect has released Prime Agent, an open-source research and coding tool that allows models to operate through sub-agent calls managed within a persistent IPython kernel. By utilizing a recursive language model architecture, the system enables agents to perform functions, modify their own prompts, and refine their operational skills during execution. This approach aims to improve task automation by maintaining a continuous, self-improving environment for complex research workflows.