Fixing a Deep Search Agent with HALO

In this video, I’ll show you how to use HALO to find and fix problems in a deep search agent. We’ll analyze real agent traces, review HALO’s findings, inspect the evidence behind its recommendations, and use a coding agent to apply those fixes directly to the codebase. The example uses a fully instrumented TypeScript deep search agent built with the OpenAI Agents SDK, along with a pre-generated trace dataset so you can follow the same workflow yourself. This video uses the hosted version of HALO on inference.net, but HALO is also fully open source and can be run locally with the free desktop app. Resources: Guide: https://docs.inference.net/guides/try... Typescript Demo Repository: https://github.com/context-labs/Searc... Python Demo Repository: https://github.com/context-labs/Searc... HALO: https://github.com/context-labs/halo Inference: https://inference.net