GitHub Copilot for JetBrains Adds MCP Agents and Telemetry Controls
The latest plugin release brings custom tools into Claude agent flows while adding observability, token-limit and model-management settings.
GitHub released a Copilot update for JetBrains IDEs on July 27 that expands both agent customization and enterprise oversight. Developers can connect MCP servers and custom agents inside Claude agent flows, while administrators and advanced users gain new telemetry, token-limit and model-management controls. The release brings more of the configurable agent stack directly into the IDE rather than leaving those controls to a separate command-line workflow.
What changed
The plugin can now export OpenTelemetry data for agent workflows. GitHub places the configuration under the Copilot chat settings and frames it as a way to align agent activity with an organization's observability requirements. The changelog does not claim that telemetry is enabled automatically, so teams still need to configure the export path that fits their monitoring environment.
Model controls also became more explicit. Users working with bring-your-own-key or custom endpoints can set default maximum input and output token limits. Built-in Copilot models can be enabled or disabled through the model-management controls. Together, those settings give teams a way to constrain request size and decide which hosted models appear in the JetBrains experience.
The agent layer now accepts MCP servers and custom agents in Claude flows. That makes specialized tools, repository-specific instructions and team workflows available without leaving the IDE. GitHub also expanded Copilot CLI sessions with forks, a rubber-duck command for reasoning through an implementation, and a visible todo list. The release includes smaller changes to model pickers, MCP diagnostics, session prompts and path handling.
Why it matters
Coding agents are becoming collections of models, tools and policies rather than a single chat feature. Putting MCP connections, custom agents, telemetry and token limits in one IDE surface reduces the distance between experimentation and governed team use. It also increases the importance of configuration review: a tool connection or custom endpoint can widen what an agent may access, even when the interface makes setup easier.
GitHub's changelog establishes availability and configuration, not performance. It does not provide independent evidence that the new controls improve code quality, reduce cost or prevent unsafe tool use. Teams should treat the release as added operational capability and test the resulting workflows against their own security and observability requirements.
Status
Confirmed. GitHub documented the features and directs users to the latest Copilot plugin for JetBrains IDEs. Internal confidence is high because GitHub is the primary authority for its plugin release; the practical benefits remain to be measured by users.
Sources
Update note: Last reviewed 2026-07-29. We will revise this post if GitHub changes plugin availability, supported agent flows or the scope of its management controls.
Sources
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.