Chainguard launches AI-powered Harness to identify and fix software vulnerabilities

Chainguard has launched Chainguard Harness, a new AI-powered cybersecurity solution designed to identify exploitable vulnerabilities in application code and provide verified fixes.

Announced on 14 October 2026, the offering uses advanced AI models, including Anthropic’s Claude Mythos 5.1, to test applications for security weaknesses. The system aims to help organisations detect and address vulnerabilities without requiring direct access to frontier AI models.

Unlike traditional security scanners that analyse static code, Chainguard Harness builds an application from its source code within an isolated virtual environment. AI models then test the running application to identify vulnerabilities, including logic flaws, injection risks, and resource exhaustion issues.

Each potential vulnerability is reproduced in a separate environment to verify the finding before a recommended fix is delivered. According to Chainguard, the process is designed to reduce false positives and give security teams actionable results.

“Every time we’ve pointed Harness at code, it has found serious, exploitable vulnerabilities, including in software that’s been scanned many times over and code that runs critical infrastructure,” said Dan Lorenc, CEO and Co-founder of Chainguard.

“That capability is beginning to make its way to more and more organizations, and the teams that move first will be the ones that find and fix these issues before they become incidents. Harness offers another option for security teams to get proven vulnerabilities and fixes their engineers can ship.”

Bringing AI-Powered Security Testing to Enterprises

Chainguard Harness connects to software repositories through the company’s Guardener GitHub application, allowing security testing to take place in isolated microVM environments.

The company states that these environments operate without network egress or stored credentials, with audit logging incorporated into the testing process.

The offering also includes human review of recommended fixes by Chainguard engineers before results are delivered to customers.

Chainguard Harness is available through Anthropic’s Claude Marketplace, allowing eligible enterprise customers to purchase the service through their existing Anthropic arrangements.

Access to the service is currently managed through a beta waitlist.

The launch reflects the increasing use of advanced AI models in enterprise cybersecurity, particularly in identifying software vulnerabilities and supporting remediation efforts.

To learn more visit Chainguard’s website.