Show HN: Lok – Treating LLMs more like infrastructure, not chatbots (github.com)

Lok, a new orchestration layer designed to manage multiple large language models (LLMs) in software development workflows, was recently introduced in a "Show HN" post. This tool acts as an intermediary that coordinates various specialized LLM backends—such as Claude Code, Gemini CLI, and Codex—to automatically analyze code, identify bugs, and propose fixes. By utilizing an agentic LLM as the conductor, Lok streamlines processes that include fetching issues from repositories, synthesizing consensus proposals, and posting comments directly to GitHub issues. Significant for the AI/ML community, Lok exemplifies a shift in how LLMs can be leveraged beyond traditional chatbot functionalities, embedding them more deeply into the software development lifecycle. With features like multi-backend bug hunting and automated PR reviews, Lok optimizes coding practices and enhances collaboration. Developers can utilize commands such as `lok hunt` for bug detection or `lok run fix` to propose fixes, reflecting a sophisticated integration of AI into coding environments. The tool's ability to support custom workflows, which can evolve based on previous steps, underscores its potential to redefine development practices and improve productivity within teams.