GitHub Copilot for JetBrains Adds More BYOK and Agent Customization Options

GitHub expanded Copilot for JetBrains with additional bring-your-own-key capabilities and new controls for agent-based development workflows, according to a July 14 GitHub Changelog post.

What changed in JetBrains

GitHub said Copilot for JetBrains now supports OpenAI-compatible custom endpoints. That matters for teams that want to connect Copilot workflows to model endpoints outside the default configuration, while staying within the JetBrains development environment.

The same GitHub Changelog entry says the JetBrains update also adds plugin management and Claude agent provider customizations for agents, skills, and instructions. GitHub also listed local sandboxing and a built-in debugger skill for Copilot CLI sessions as part of the release.

The changelog does not describe the update as a general release of every feature to every Copilot customer. Based on GitHub’s wording, the main change is an expansion of configuration and customization options inside Copilot for JetBrains, especially around custom endpoints and agent-related behavior.

Broader Copilot updates around security

The JetBrains update arrived alongside several other Copilot changes announced by GitHub in July.

In a separate July 14 GitHub Changelog post, GitHub announced that security reviews are available in the GitHub Copilot app through a public-preview /security-review slash command. GitHub said the command scans in-flight code changes for high-confidence vulnerabilities and provides actionable fixes inside the coding workflow.

GitHub’s description positions the feature as a way to review changes before they are merged or shipped, rather than as a replacement for a full security program. The company’s claim is narrower: Copilot can analyze active code changes and return suggested fixes for vulnerabilities it identifies with high confidence.

On July 10, GitHub also announced agentic autofix for code scanning alerts in public preview. According to GitHub Changelog, the feature can be assigned code scanning alerts, explore a codebase, generate and validate fixes with CodeQL, and open draft pull requests through the Copilot cloud agent.

That update connects Copilot more directly to GitHub’s code-scanning workflow. GitHub said CodeQL is used to validate fixes, and the final output can be a draft pull request rather than only an inline suggestion.

VS Code Copilot also gained workflow features

GitHub’s July 8 Changelog roundup for Copilot in Visual Studio Code described several additional capabilities released in June 2026. GitHub said those updates included agentic browser tools, parallel agent sessions, subagent cost visibility, and managed settings.

The VS Code roundup also said Copilot supports 1 million-token context windows for compatible Anthropic and OpenAI models. GitHub separately noted MCP OAuth credentials in the same roundup.

Taken together, the recent changelog posts show GitHub continuing to add configuration, security, and automation features across Copilot’s editor and app surfaces. The JetBrains-specific news is most relevant for developers using JetBrains IDEs who need custom model endpoint support, more control over plugins, or more granular customization of agent behavior.