Portable Conductor Plugin Brings Spec-Driven Dev to Claude
Google has decoupled Conductor from the Gemini CLI, releasing a portable plugin that brings conversational Spec-Driven Development to Antigravity and Claude.
On July 16, 2026, Google Developers transitioned Conductor from a dedicated Gemini CLI extension into a portable plugin architecture. The official release expands Spec-Driven Development (SDD) across external ecosystems, most notably adding native integration for the Antigravity CLI and Anthropic’s Claude model family. The architectural shift signals a broader move by Google toward model-agnostic orchestration for development workflows.
Portable Plugin Architecture
The primary technical change decouples Conductor’s core logic from Google’s proprietary toolchain. Conductor now functions as a modular engine designed to be injected into third-party command-line environments.
The initial rollout centers on Antigravity, a high-performance developer toolchain. Developers can now manage complex project specifications and trigger code generation directly through the Antigravity interface while routing the underlying reasoning tasks through Conductor.
By unbundling the orchestration layer from the model execution, Google allows developers to automate workflows using the models best suited for specific engineering tasks. The official inclusion of Claude indicates a pragmatic acknowledgment that multi-model environments are the standard for modern software engineering.
Conversational Spec-Driven Development
Traditional scaffolding tools rely on rigid command sequences for state changes. Conductor replaces this paradigm with a conversational interface that dynamically manages project state.
As you chat with the CLI, Conductor maintains a hidden specification file, or “Spec”, updating implementation details and requirements in real-time. When switching between tasks, the engine injects this current Spec directly into the model’s context window. This ensures the model remains grounded in the project’s specific technical constraints without requiring manual prompt engineering.
The Antigravity integration builds upon this with automated validation loops. The CLI evaluates all generated code against the active Spec before presenting it to the user, reducing the risk of drift in long-running agent interactions. This approach shifts the developer’s role from managing individual syntax tasks to governing the overarching system state, similar to how developers evaluate and test AI agents in production.
Capability Shift
The transition reshapes the constraints of the Conductor environment:
| Feature | Legacy Conductor | Portable Conductor Plugin |
|---|---|---|
| Core Architecture | Gemini CLI Extension | Modular CLI Plugin |
| Supported Models | Gemini | Gemini, Claude |
| Toolchain Integration | Google-specific | Antigravity, External CLI tools |
| State Management | Command-driven scaffolding | Conversational real-time spec updates |
Installation and Availability
The portable plugin format is available immediately via the Google Developers GitHub repository. Developers operating within the Antigravity environment can install the update using the standard antigravity plugin add conductor command.
If you manage complex codebases using Spec-Driven Development, the decoupling of Conductor allows you to centralize your architectural requirements in a persistent state file while leveraging Claude for execution. You can now rely on the CLI to handle context injection and validation automatically.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
How to Configure Elastic Training in MaxText on TPUs
Learn how to enable elastic training in MaxText to survive hardware failures and resume distributed AI workloads in seconds.
Google Vertex Adds Mirendil's Self-Evolving RFT Architecture
Mirendil secured a $100 million Google Cloud partnership to scale its self-improving RFT architecture to 50 trillion tokens and integrate with Vertex AI.
Google Cloud API Gateway Adds Virtual Routing for AI Models
Google Cloud's new public preview brings in-flight prompt transcoding and virtual model mapping to API Gateway for seamless failover across AI providers.
Unified Evaluation Engine Hits Gemini Enterprise Platform
Google Cloud has released a unified evaluation engine for the Gemini Enterprise Agent Platform to consistently measure agent quality across environments.
Gemini API Adds Tool Hooks and 3.6 Flash to Managed Agents
Google updated its Gemini API Managed Agents with the faster 3.6 Flash model and new environment hooks for validating remote tool execution.