JFrog Codex Plugin

OpenAI and JFrog have partnered to connect Codex to the JFrog Software Supply Chain Platform. Because Codex works asynchronously and reports back when finished, the plugin places security context at the moment the agent resolves a dependency rather than at review, so delegated work returns already aligned with policy.

Codex Plugin Features

JFrog Artifactory Integration

Codex works with your repositories, builds, projects, permissions, and release lifecycle through natural language, backed by your live platform instance. Security audits, CVE lookups, and Advanced Security exposure queries are available while the agent works a task, not after it finishes.

Package Safety Checks Before Download

Before the agent pulls a dependency, the JFrog package safety skill checks it against Public Catalog data and your curation policies, resolving through Artifactory remote caches rather than public registries. The check happens at resolution — the point where blocking a bad package is cheap.

Governed MCP Server Management with Agent Guard

Codex discovers, installs, and removes MCP servers only from what your project has approved in the JFrog AI Catalog, with credentials held by the platform rather than exposed in agent context.

AI Asset and Platform Administration

The JFrog AI Catalog skill surfaces approved models and AI assets to the agent, and plugin capabilities extend into broader platform administration workflows across the software development lifecycle.

About Open AI (Codex)

OpenAI is an AI research and deployment company. Codex, its agentic coding tool, is built to take a task, work on it asynchronously, and return a result — a branch, a fix, a refactor, or a set of changes ready for review.