Salt Security launches Salt Code for policy enforcement inside AI coding assistants

Salt Security announced Salt Code, a component of its Agentic Security Platform, to enforce security policies inside AI coding assistants used in software development.

What Salt Security announced

In a press release, Salt Security described Salt Code as an “agentic security” solution for AI-generated code workflows. The company says the product is designed to apply security controls where developers are increasingly using AI coding assistants to write, revise and review code.

According to Salt Security’s announcement, Salt Code supports a range of assistants and developer tools, including Claude Code, Cursor, GitHub Copilot, Windsurf, Kiro, Codex, Gemini CLI and Antigravity. Salt Security’s product page also lists support for Claude Code, Cursor, GitHub Copilot, Codex and Gemini CLI, and says Salt Code connects the company’s Salt Posture Governance Engine to developer tools.

Salt Security positions the launch as a way for organizations to govern AI-assisted development without relying only on downstream checks after code has already been produced. The company says Salt Code enforces policies inside coding assistants, which suggests the product is intended to bring security requirements closer to the point where code is generated or modified.

Policy packs and early access

Salt Security’s access page says Salt Code is available in early access with four secure coding packs and more than 40 policies. The same page says the product supports Cursor, GitHub Copilot, Claude and assistants that support MCP, a protocol used to connect AI tools with external systems and context.

The company also says Salt Code is “free to start” and available at no cost to the first 100 organizations. Salt Security’s access page frames the initial offering around secure coding packs, rather than a broad replacement for existing application security testing tools.

Why it matters

AI coding assistants are now used in development environments where code may be produced faster than traditional review processes can keep up. Salt Security’s announcement reflects a growing enterprise concern: teams want the productivity benefits of AI-assisted development while still applying internal rules for secure coding, data handling and software governance.

The notable part of Salt Code is its focus on policy enforcement inside the assistant workflow. Many security tools scan repositories, dependencies or running applications after code is written. Salt Security’s materials instead emphasize integration with the coding assistant layer, where a developer may be prompted by, or collaborating with, an AI system.

That approach could be relevant for security teams trying to standardize guardrails across multiple assistants. Salt Security lists support for several commonly used AI coding tools, including GitHub Copilot, Cursor and Claude Code. If teams use more than one assistant, centralized policy enforcement may be attractive, though Salt Security’s public materials do not provide independent performance benchmarks or detailed deployment results.

What remains unclear

Salt Security’s public pages describe the supported assistants, policy packs and connection to its governance engine, but they do not include detailed third-party evaluations in the provided materials. The sources also do not specify how Salt Code handles every type of policy violation, how it integrates with existing CI/CD controls, or what administrative capabilities are included beyond the early-access description.

For now, Salt Code should be understood as Salt Security’s new product for bringing security policy enforcement into AI-assisted coding environments. The launch adds to a broader shift in security tooling, where vendors are adapting controls for developers who increasingly work with AI assistants during day-to-day coding.