Andrii Bidochko

• Updated: February 2, 2026
• 3 min read

Google Conductor: A Context‑Driven Gemini CLI Extension Transforming AI Agent Development

Google Conductor: A Context‑Driven Gemini CLI Extension Transforming AI Agent Development

Meta Description: Discover how Google Conductor, the new Gemini CLI extension, leverages Markdown‑based context storage to streamline AI agent workflows. Learn its features, installation steps, and why it matters for developers building intelligent agents.

What Is Google Conductor?

Google Conductor is an open‑source extension for the Gemini Command‑Line Interface (CLI) that introduces a context‑driven development model for AI agents. By persisting project knowledge in version‑controlled Markdown files, Conductor enables repeatable, auditable, and collaborative AI development workflows.

Key Features and Benefits

• Markdown‑Based Knowledge Base: All project specifications, prompts, and implementation notes are stored as markdown, making them searchable and versioned.
• Track‑Oriented Workflow: Projects are organized into tracks – each with a spec.md (desired outcome) and a plan.md (step‑by‑step execution plan).
• CLI Commands for Full Lifecycle:
  • conductor init – bootstrap a new track.
  • conductor implement – run the plan and generate outputs.
  • conductor status – view current progress.
  • conductor review – compare generated results with expectations.
  • conductor rollback – revert to a previous version if needed.
• Versioned Context: Every change creates a new commit, allowing developers to trace how prompts evolved over time.
• Seamless Integration: Works natively with Gemini models, making it easy to plug into existing AI pipelines.

Installation Steps

1. Ensure you have the latest Gemini CLI installed.
2. Run gemini extension install conductor to add the Conductor plugin.
3. Initialize a new project with conductor init my‑project. This creates a .conductor folder with default markdown templates.
4. Edit spec.md to describe the AI agent’s goal and plan.md to outline the steps.
5. Execute conductor implement to let Gemini generate the agent based on the plan.

How the Workflow Looks in Practice

Imagine building a customer‑support chatbot. You would:

• Write a spec.md that defines the bot’s persona, knowledge base, and success criteria.
• Draft a plan.md listing tasks such as data ingestion, prompt engineering, and testing.
• Run conductor implement – Gemini reads the plan, executes each step, and stores outputs (code snippets, prompts, test results) back into markdown files.
• Use conductor review to compare the generated bot’s behavior against the spec, iterating until the desired performance is reached.

This loop ensures that every decision is documented, reproducible, and reversible.

Impact on AI Agent Development

Conductor shifts AI development from ad‑hoc script‑hacking to a structured, version‑controlled process. Benefits include:

• Transparency: Stakeholders can read the markdown spec to understand what the agent is expected to do.
• Collaboration: Teams can contribute via pull requests, just like traditional software.
• Auditability: Historical context is preserved, simplifying compliance and debugging.
• Scalability: Re‑using tracks across projects accelerates time‑to‑market for new agents.

Why It Matters for UBOS Users

UBOS’s AI ecosystem already supports Gemini‑based agents, markdown‑driven workflows, and modular extensions. Google Conductor aligns perfectly with our internal tools:

• Explore more about AI agents on our AI Agents hub.
• Learn how to integrate Gemini CLI extensions at UBOS Gemini CLI guide.
• Discover best practices for markdown‑centric development on Markdown Workflows.

Get Started Today

The Conductor repository is open on GitHub, complete with documentation, sample tracks, and a community forum. Dive in, clone a starter track, and experience a new level of control over your AI agents.

Visit the Google Conductor GitHub repo | Read the original announcement

Ready to accelerate your AI projects? Explore Conductor, experiment with our UBOS resources, and build the next generation of context‑aware agents.

Andrii Bidochko

CTO UBOS

Andrii Bidochko is an AI entrepreneur and researcher focused on AI agents, reinforcement learning, and autonomous systems. He writes about the technologies shaping the future of machine intelligence, from frontier models and agent architectures to real-world AI applications.