# Build What’s Next on the ChatGPT Platform with Plugins | DevDay 2026

## Executive summary

OpenAI announced a major expansion of the ChatGPT platform for developers, focusing on 'plugins' and 'plugin extensions.' These updates allow external tools and services to integrate deeply into the ChatGPT interface, enabling complex, multi-step workflows. Key improvements include enhanced plugin discovery via conversational suggestions, native sidebar/side panel integration for a more collaborative user experience, and significant overhauls to the plugin submission and deployment process, making it easier for developers to build and scale full-stack applications using the MCP framework.

## Key takeaways

- Enhanced Plugin Discovery: Plugins can now be suggested directly within a conversation when ChatGPT identifies that a user's task requires external tools. Discovery is shaped by three signals: Conversion (usage after installation), Active Usage (daily/weekly use), and Retention (repeat usage).
- Native Plugin Extensions: Developers can bring their product interfaces into the ChatGPT sidebar or side panel, allowing for a highly native, interactive experience. This includes the ability to share context (e.g., an element selection in the side panel) back to the model and the conversation.
- Simplified Build and Shipping Workflow: The plugin submission process has been overhauled: the submission form is eliminated, and required metadata is generated from the `plugin.json` file. Review checks are now automated upon upload, and MCP server updates no longer require resubmission.

## Technical details

- Multi-Context Processing (MCP) & Agents: MCP is becoming the core mechanism for how agents interact with the world, enabling complex tasks that involve multiple tool calls and context pulling. The platform is introducing support for the upcoming MCV event spec, allowing the MCP server to notify ChatGPT of events and initiate automated workflows via `event/list` and `event/subscribe` methods.
- Plugin Extension Implementation: Extensions can be integrated into the ChatGPT sidebar by adding `entry points` configuration to the MCP tool. Developers can also register file entry points to allow their plugin to open specific file types interactively within the side panel. Furthermore, plugins can expose read-only MCP tools marked for `mentioned search` within the composer.
- User Onboarding and Settings: Plugins can now provide a custom onboarding skill that guides users through setup. Configuration settings can be exposed directly in the ChatGPT settings screen by providing a read tool (schema/current values) and an update tool.
- Hosting and Identity: The platform supports hosting MCP servers via 'sites,' enabling full-stack web apps. Developers can increase plugin usage by implementing 'Sign In with ChatGPT' (an identity integration), allowing users to bring their ChatGPT token entitlements into the product.

## Practical implications

- Developers should focus on building experiences that are highly interactive and collaborative, leveraging the side panel for visual data inspection and modification.
- The shift toward automated review and deployment means developers must ensure their `plugin.json` is comprehensive and that their MCP server adheres to the new event and entry point specifications.
- Implementing 'Sign In with ChatGPT' is a recommended strategy to reduce user friction and increase plugin adoption.

## Topics

Build Engineering, AI Agents, Plugin Architecture, API Integration, Developer Workflow, OpenAI developers plugins, App Plugin Creator, Codex

Source: https://www.youtube.com/watch?v=pIW_8VyvCuQ
