Agent Runners

Product Introduction

1. Agent Runners is a Netlify platform feature that enables teams to deploy AI-generated code directly from the Netlify Dashboard using preconfigured AI agents like Claude Code, Codex, or Gemini. It integrates AI-generated code into live production environments without requiring manual setup or infrastructure configuration. The product streamlines the process of turning AI outputs into deployable code with built-in previews, Git compatibility, and audit controls.
2. The core value of Agent Runners lies in bridging the gap between AI-generated code prototypes and production-ready deployments. It accelerates development cycles by allowing teams to test, refine, and ship AI-driven changes in real production contexts while maintaining enterprise-grade security and workflow compliance.

Main Features

1. Agent Runners provide instant deployment of AI-generated code with live previews, enabling users to validate changes in real-time before merging into production. Every code iteration generates a shareable URL for collaborative review, ensuring visibility across teams.
2. The product supports multiple AI agents (Claude Code, Codex, Gemini) within the Netlify Dashboard, eliminating the need for API key management or external tool integration. Agents operate with full access to the application’s production context, including runtime environment variables and dependencies.
3. Enterprise-grade safeguards include isolated execution environments, granular role-based access controls (RBAC), and automatic version history tracking. All deployments are reversible, and audit logs capture user prompts, AI actions, and deployment timelines for compliance.

Problems Solved

1. Agent Runners address the disconnect between AI-generated code prototypes and production infrastructure, which often causes deployment failures or environment mismatches. By testing changes directly in live environments, teams avoid staging-only validation gaps.
2. The product targets cross-functional teams, including developers, product managers, and designers, who need to collaborate on rapid iterations without deep technical setup. It particularly benefits organizations managing technical debt or backlogged feature requests.
3. Typical use cases include patching broken production links, resolving runtime errors identified in logs, prototyping UX/UI adjustments, and implementing backend logic changes suggested by AI agents. Non-technical users can safely contribute code updates via guided prompts.

Unique Advantages

1. Unlike standalone AI coding tools, Agent Runners embed AI workflows directly into Netlify’s deployment pipeline, providing real-time access to production data, environment variables, and team-specific Git repositories. This eliminates manual context-sharing with AI models.
2. The product innovates with sandboxed execution environments that mirror production without exposing secrets, combined with automatic pull request generation for GitHub-integrated workflows. Changes are tracked as Git commits, aligning with existing developer processes.
3. Competitive differentiation stems from Netlify’s integrated platform: AI deployments inherit built-in DDoS protection, global CDN distribution, and compliance certifications (SOC 2, GDPR). No other AI-code tool offers equivalent end-to-end deployment safeguards.

Frequently Asked Questions (FAQ)

1. How does Agent Runners integrate with existing GitHub workflows? Agent Runners automatically generates pull requests for AI-driven code changes, allowing teams to enforce code reviews or CI/CD checks before merging. Deployments sync with your repository’s branch structure, maintaining version control parity.
2. Which AI models are supported, and can custom models be added? The platform currently supports Claude Code, Codex, and Gemini out of the box. Custom model integration is not available, but Netlify plans to expand the agent library based on user demand.
3. How are security and permissions managed for non-technical users? RBAC limits AI actions based on team roles: users without deployment rights can generate previews but cannot merge to production. All prompts and code outputs are logged, and secrets are never exposed to AI agents.