Rubrik releases Code Guardian and adds Model Context Protocol support

Updated 19 Sep 2026 · Reviewed by the Zekai editorial team

Security vendor Rubrik has launched a new product called Code Guardian, designed to review source code for complex security flaws. According to the company, the service operates on an air-gapped copy of a customer's codebase rather than scanning a live repository directly. This approach aims to reduce potential risks to production systems while examining files, services, and cloud environments for multi-step vulnerability chains.

The product is built around Anthropic's Claude Mythos 5 model. Michael Moore, a cybersecurity lead at Anthropic, claims the integration allows defenders to validate attack paths before malicious actors can exploit them. Rubrik states that once confirmed critical issues are found, the service can route them directly into tools like Jira or GitHub for remediation, while also linking to recovery workflows.

Alongside the code review launch, Rubrik introduced support for the Model Context Protocol. This open standard is intended to let AI models and agents access external tools and data more consistently. By exposing the schema of its Security Cloud APIs, the company enables connected agents to query available functions, letting teams save compliance or recovery processes as reusable utilities.

Arvind Nithrakashyap, Rubrik's Chief Technology Officer, states that the company developed the internal software harness using frontier AI before offering it to customers. The firm notes that its AI agent architecture was built alongside Anthropic to improve multi-step reasoning and reduce response latency during incident response, though the current agent features are also used for administrative tasks like summarizing overnight backup failures.

For developers and security engineers, these updates represent a shift toward using generative AI for deeper technical analysis rather than just alert summaries. Code Guardian is presently accepting select design partners in a private preview phase. Meanwhile, the Model Context Protocol integration is available in a private preview exclusively for existing Rubrik customers.

Why it matters for AI for Software Development

For software developers and security engineers, these releases indicate how frontier models are being integrated into internal pipelines to trace complex, multi-file vulnerability chains. By utilizing isolated code copies and connecting AI agents through standard protocols like MCP, development teams can automate aspects of threat detection and remediation workflows without granting unvetted access to live production systems.

What to do about it

• Evaluate whether your organization qualifies for the private preview of Code Guardian for isolated source code analysis.
• Examine the Model Context Protocol documentation to understand how to connect custom AI agents to security schemas.
• Review how verified vulnerabilities from automated scanning tools integrate with your existing Jira and GitHub workflows.

Tools mentioned

• GitHub
• Claude
• Jira

Source: Rubrik launches Code Guardian & expands Anthropic ties – SecurityBrief Australia (securitybrief.com.au)