JFrog today at its swampUP 2026 conference added a zero touch remediation capability that ensures the most secure version of a binary is provided even when application developers request a version that has known vulnerabilities.
Additionally, JFrog is adding tools and capabilities to secure artificial intelligence (AI) agents that have been embedded within a DevOps workflow. For example, an AI Asset Scanning tool uses semantic scanning of markdown files, skills scripts, and instruction sets found in AI models, skills, plugins and Model Context Protocol (MCP) servers to block malicious behavior.
There is also now an Agent Guard that extends the company’s core Artifactory registry to apply governance policies to plug-ins for AI coding agents. It natively enforces project-scoped allow/deny policies from AI Catalog within developer tools to ensure coding agents never bypass organizational guidelines and only consume approved AI assets.
JFrog also revealed today it has added support for Agent Package Manager (APM), a dependency manager for AI agents developed by Microsoft within its registry. APM authenticates AI agents and then creates a trusted path for agents to resolve package dependencies using the JFrog Artifactory in a way that ensures only trusted, verified, audited components are used to build applications. JFrog is also making available JFrog Traffic Controller to block public registry calls at the network layer, rerouting instead all traffic through Artifactory. Cloudflare, Netskope and Zscaler also today announced support for JFrog Traffic Controller.
At the same time, JFrog also announced today an alliance with the Wiz unit of Google Cloud through which DevSecOps teams will be provided with a unified view of what’s running, where it came from, whether it’s trusted, and how to fix it.
Finally, JFrog is extending JFrog AppTrust, a framework for enforcing policies across a software development lifecycle (SDLC) to enable DevSecOps teams to create policies-as-code in plain English using an AI tool. There is also a prompt-to-release traceability capability that tracks approvals, builds, scans, and promotions without any manual logging required. JFrog is also now providing compliance templates with pre-built rules aligned with regulatory standards that are automatically enforced with one click and a monitoring capability that surfaces compliance issues after an application has been deployed.
JFrog CTO Yoav Landman told conference attendees that collectively these capabilities make it possible for JFrog to immunize artifacts by limiting what AI agents can access, heal any issues that might arise and then govern how artifacts are deployed. The overall goal is to identify vulnerabilities and remediate them as soon as a fix is available without requiring any human intervention in a way that prevents builds from being broken.
That goal is achieved because JFrog Curation with Compliant Version Selection now blocks risky packages, AI assets, IDE extensions, and third-party components the moment a developer tries to pull them. JFrog Xray and JFrog Advanced Security then scan release artifacts for every dependency to catch vulnerable code patterns before they ship and surface exposed credentials across every artifact type. JFrog Contextual Analysis adds the ability to evaluate every finding for reachability and exploitability, while JFrog Runtime narrows what is actually loaded in production.
It’s not clear at what rate DevSecOps teams are more proactively moving to lock software supply chains but it’s apparent that the volume of AI-enabled attacks against them is only going to increase. The issue now is determining how best to secure software supply chains before what is rapidly becoming a tsunami of application vulnerabilities are exploited more rapidly than they already are.
from DevOps.com https://ift.tt/n4oKi8M
Comments
Post a Comment