One common practice for working with MCP tools calls is to put the outputs from a tool back into the LLM as a message, and ask the LLM for the next step. ...| jngiam
Open source “Deep Research” project proves that agent frameworks boost AI model capability.| Ars Technica