Google Releases Conductor: a context pushed Gemini CLI extension that shops information as Markdown and orchestrates agentic workflows

Google has launched Conductor, an open supply preview extension for Gemini CLI that turns AI code era right into a structured, context pushed workflow. Conductor shops product information, technical choices, and work plans as versioned Markdown contained in the repository, then drives Gemini brokers from these recordsdata as an alternative of advert hoc chat prompts.

From chat primarily based coding to context pushed growth

Most AI coding right now is session primarily based. You paste code right into a chat, describe the duty, and the context disappears when the session ends. Conductor treats that as a core downside.

As a substitute of ephemeral prompts, Conductor maintains a persistent context listing contained in the repo. It captures product objectives, constraints, tech stack, workflow guidelines, and elegance guides as Markdown. Gemini then reads these recordsdata on each run. This makes AI conduct repeatable throughout machines, shells, and workforce members.

Conductor additionally enforces a easy lifecycle:

Context → Spec and Plan → Implement

The extension doesn’t leap instantly from a pure language request to code edits. It first creates a observe, writes a spec, generates a plan, and solely then executes.

Putting in Conductor into Gemini CLI

Conductor runs as a Gemini CLI extension. Set up is one command:

gemini extensions set up https://github.com/gemini-cli-extensions/conductor --auto-update

The --auto-update flag is non-compulsory and retains the extension synchronized with the newest launch. After set up, Conductor instructions can be found inside Gemini CLI when you’re in a undertaking listing.

Challenge setup with /conductor:setup

The workflow begins with undertaking degree setup:

This command runs an interactive session that builds the bottom context. Conductor asks concerning the product, customers, necessities, tech stack, and growth practices. From these solutions it generates a conductor/ listing with a number of recordsdata, for instance:

• conductor/product.md
• conductor/product-guidelines.md
• conductor/tech-stack.md
• conductor/workflow.md
• conductor/code_styleguides/
• conductor/tracks.md

These artifacts outline how the AI ought to motive concerning the undertaking. They describe the goal customers, excessive degree options, accepted applied sciences, testing expectations, and coding conventions. They stay in Git with the remainder of the supply code, so adjustments to context are reviewable and auditable.

Tracks: spec and plan as top notch artifacts

Conductor introduces tracks to symbolize items of labor similar to options or bug fixes. You create a observe with:

or with a brief description:

/conductor:newTrack "Add darkish mode toggle to settings web page"

For every new observe, Conductor creates a listing below conductor/tracks// containing:

• spec.md
• plan.md
• metadata.json

spec.md holds the detailed necessities and constraints for the observe. plan.md accommodates a stepwise execution plan damaged into phases, duties, and subtasks. metadata.json shops identifiers and standing data.

Conductor helps draft spec and plan utilizing the prevailing context recordsdata. The developer then edits and approves them. The necessary level is that every one implementation should comply with a plan that’s express and model managed.

Implementation with /conductor:implement

As soon as the plan is prepared, you hand management to the agent:

Conductor reads plan.md, selects the subsequent pending process, and runs the configured workflow. Typical cycles embody:

1. Examine related recordsdata and context.
2. Suggest code adjustments.
3. Run assessments or checks in response to conductor/workflow.md.
4. Replace process standing in plan.md and international tracks.md.

The extension additionally inserts checkpoints at section boundaries. At these factors Conductor pauses for human verification earlier than persevering with. This retains the agent from making use of giant, unreviewed refactors.

A number of operational instructions help this circulate:

• /conductor:standing reveals observe and process progress.
• /conductor:evaluation helps validate accomplished work in opposition to product and elegance tips.
• /conductor:revert makes use of Git to roll again a observe, section, or process.

Reverts are outlined by way of tracks, not uncooked commit hashes, which is less complicated to motive about in a multi change workflow.

Brownfield tasks and workforce workflows

Conductor is designed to work on brownfield codebases, not solely contemporary tasks. Whenever you run /conductor:setup in an present repository, the context session turns into a option to extract implicit information from the workforce into express Markdown. Over time, as extra tracks run, the context listing turns into a compact illustration of the system’s structure and constraints.

Crew degree conduct is encoded in workflow.md, tech-stack.md, and elegance information recordsdata. Any engineer or AI agent that makes use of Conductor in that repo inherits the identical guidelines. That is helpful for imposing take a look at methods, linting expectations, or authorized frameworks throughout contributors.

As a result of context and plans are in Git, they are often code reviewed, mentioned, and altered with the identical course of as supply recordsdata.

Key Takeaways

• Conductor is a Gemini CLI extension for context-driven growth: It’s an open supply, Apache 2.0 licensed extension that runs inside Gemini CLI and drives AI brokers from repository-local Markdown context as an alternative of advert hoc prompts.
• Challenge context is saved as versioned Markdown below conductor/: Recordsdata like product.md, tech-stack.md, workflow.md, and code model guides outline product objectives, tech selections, and workflow guidelines that the agent reads on every run.
• Work is organized into tracks with spec.md and plan.md: /conductor:newTrack creates a observe listing containing spec.md, plan.md, and metadata.json, making necessities and execution plans express, reviewable, and tied to Git.
• Implementation is managed by way of /conductor:implement and track-aware ops: The agent executes duties in response to plan.md, updates progress in tracks.md, and helps /conductor:standing, /conductor:evaluation, and /conductor:revert for progress inspection and Git-backed rollback.

Take a look at the Repo and Technical particulars. Additionally, be happy to comply with us on Twitter and don’t neglect to affix our 100k+ ML SubReddit and Subscribe to our E-newsletter. Wait! are you on telegram? now you possibly can be a part of us on telegram as effectively.