source: kdnuggets: getting started with conductor for gemini cli
level: technical
conductor, released in preview on december 17, 2025, is a gemini cli extension that introduces context-driven development. instead of describing your project fresh every session, you maintain markdown files in a conductor/ directory that define your product vision, tech stack, coding standards, and feature plans. the agent reads these files on every run, so your context persists across sessions. this fixes the common problem where ai coding agents make plausible but wrong guesses because they lack project knowledge.
the workflow uses six slash commands. /conductor:setup analyzes your codebase or guides you through creating context artifacts like product.md, tech-stack.md, and code style guides. /conductor:newtrack creates a spec and phased implementation plan for a feature. /conductor:implement works through the plan task by task, committing after each one and pausing for manual verification at phase boundaries. supporting commands include status, revert, and review. the system works for both new and existing projects, and it uses git for per-task commits and rollback.
to get started, install gemini cli globally, set up authentication with a google api key or vertex ai, and initialize git in your project. then install the conductor extension from its github repository. run /conductor:setup to build your project context, then use /conductor:newtrack to start a feature and /conductor:implement to execute it. the state is saved to disk, so you can stop and resume anytime. the conductor/ directory can be committed to share context with teammates.
why it matters: it makes ai coding agents reliable by giving them permanent project context, reducing wasted time on incorrect code and manual corrections.
source: kdnuggets: getting started with conductor for gemini cli