OpenAI rolls out Codex plugins to connect Slack, Figma, Notion, Gmail, and more

What OpenAIDevs posted on X

On March 26, 2026, OpenAI's developer account said plugins are rolling out in Codex and framed the update around practical tool access, naming Slack, Figma, Notion, Gmail, and more. The signal here is not just another connector list. OpenAI is pushing Codex beyond isolated code generation toward the planning, coordination, and handoff work that surrounds software delivery.

That matters because the friction in agentic development often appears outside the editor. Teams need an agent to read specs, pull context from docs, coordinate tasks, and move outputs into the systems they already use. A plugin layer is a direct attempt to reduce that integration tax.

What the official docs add

OpenAI's Codex documentation says plugins are installable bundles for reusable Codex workflows. The docs say a plugin can package skills, optional app integrations, and MCP server configurations in one place. In other words, OpenAI is treating plugins as a distribution format for both tool access and workflow intelligence, not as thin one-off connectors.

The docs also show how the rollout is meant to surface in daily usage. In the Codex app, curated plugins appear in the Codex directory. In the CLI, the plugin surface is exposed through /plugins. OpenAI also documents local plugin creation, which implies teams can standardize internal workflows instead of waiting only for centrally curated packages.

Why this is high-signal for AI/IT teams

The larger product implication is that Codex is being positioned as a workflow hub, not only a coding model. If skills, tool connections, and shared server context can travel together as a plugin, onboarding and reuse become much easier across teams and repositories.

For engineering organizations, that can change how agent adoption scales. Instead of every developer assembling a custom prompt stack and tool setup, a team can distribute a repeatable package with the right connectors and task instructions already wired in. That improves consistency, reduces setup time, and gives platform teams a clearer control point for governance.

It also suggests a broader shift in competitive focus. Agent products increasingly need to integrate with the systems where work already lives, not just generate strong outputs in isolation. OpenAI's plugin rollout is important because it addresses that operational layer directly.

Sources: OpenAIDevs X post · OpenAI Codex plugins docs