WPA MCP Early Preview

Windows Performance Analyzer (WPA) helps you inspect Event Tracing for Windows (ETW) traces and diagnose performance issues across CPU, memory, disk, networking, scheduling, input, and other system activity. WPA MCP is an Early Preview feature that connects WPA trace data to GitHub Copilot CLI so you can ask performance-analysis questions in natural language.

With WPA MCP, you can describe what you want to investigate instead of manually navigating every graph, table, filter, and time range yourself. GitHub Copilot CLI can help identify relevant WPA data, query trace tables, summarize findings, and suggest follow-up areas to inspect.

Important

WPA MCP is an Early Preview feature. Behavior, availability, supported scenarios, and user experience may change before general availability.

Current support

The current Early Preview release supports GitHub Copilot integration only. You need an active GitHub Copilot subscription to use WPA MCP.

Other AI assistants, model providers, and MCP clients aren't supported in this Early Preview release.

Why use WPA MCP?

Trace analysis can be powerful but difficult to scale. A single trace can include large volumes of ETW data across multiple system components, and finding the right signal often requires domain expertise and familiarity with WPA views, graphs, columns, and diagnostic workflows.

WPA MCP helps reduce that complexity by letting you:

  • Ask trace-analysis questions in plain language.
  • Find relevant WPA tables and views faster.
  • Investigate CPU usage, expensive call stacks, delays, regressions, and other performance symptoms.
  • Continue analysis with follow-up questions after initial analysis.
  • Shorten the path from trace exploration to actionable insight.

Prerequisites

Before you start, make sure you have:

  • A Windows device with a WPA version that includes the WPA MCP Early Preview feature.
  • An active GitHub Copilot subscription.
  • GitHub Copilot CLI installation configured for the preview environment.
  • A trace file that can be opened in WPA.

Note

Exact setup requirements may vary during the Early Preview period. Follow the preview release notes or onboarding instructions provided with your WPA build.

Open WPA MCP

  1. Open a trace in WPA.

  2. Select the GitHub Copilot button in the upper-right corner of the WPA window.

Initial WPA window with GitHub Copilot CLI highlighted.

  1. A Copilot pane opens on the right side of WPA.

WPA window GitHub Copilot CLI window.

  1. Enter a natural-language prompt in the Copilot pane.

    Example:

    Analyze this trace and tell me why this machine had input delays.

  2. Review the analysis. Copilot may query WPA data tables, summarize possible causes, and suggest areas to inspect.

  3. Ask follow-up questions to continue the investigation.

    Example:

    Which processes contributed most to the delay?

    Example:

    Show me the most expensive call stacks during the affected time range.

Example prompts

Use prompts that describe the symptom, time range, or question you want to investigate.

Scenario Example prompt
Input delay Analyze this trace and tell me why this machine had input delays.
CPU usage Find the processes and call stacks responsible for the highest CPU usage.
Regression investigation Compare activity before and during the slowdown and identify suspicious changes.
Disk activity What disk activity coincides with the performance issue?
Follow-up analysis What evidence supports that conclusion?

Validate Copilot results

Copilot output is generated by a large language model (LLM), and responses may vary between runs. Treat Copilot analysis as an assistant-generated starting point, not as a final diagnosis.

When using WPA MCP:

  • Validate findings against the underlying WPA trace data.
  • Review the tables, graphs, time ranges, and call stacks referenced in the analysis.
  • Use follow-up prompts to ask for supporting evidence.
  • Confirm conclusions before making engineering, product, or release decisions.

Important

LLM-generated analysis can be incomplete or incorrect. Always verify important findings in WPA before taking action.

Known limitations in Early Preview

  • Only GitHub Copilot CLI is supported in the current release.
  • A GitHub Copilot subscription is required.
  • Responses can vary between runs due to the nature of LLM.
  • Some analysis may require follow-up prompts to narrow the time range, process, thread, or subsystem.
  • Feature behavior and UI details may change during the Early Preview period.

Troubleshooting

Issue What to check
The Copilot button isn't visible Confirm that you're using a WPA build that includes WPA MCP Early Preview.
Copilot doesn't respond Confirm that GitHub Copilot CLI integration is configured and that your subscription is active.
The response is too broad Ask a more specific question that includes the symptom, time range, process, or subsystem.
The response seems incorrect Validate the result in WPA and ask Copilot for the data or evidence behind the conclusion.

Settings

With the release of WPA-MCP, we also have a set of new settings to better support for WPA MCP. You can access these Settings from the File Menu or directly from the startup launcher screen.

Window showing WPA MCP settings page.

MCP server

  • Server status – Provides information about the WPA MCP status.
  • Start with WPA – Controls whether WPA MCP starts automatically when GitHub Copilot CLI launches inside WPA. This setting is on by default. If disabled, start the server manually by entering /MCP and selecting WPA-MCP from the list of available MCP servers.
  • MCP tools – Provides read-only access to the list of tools that WPA MCP exposes to query the trace data.

Copilot CLI Settings

  • Play sound for Notifications – Simple as the title, it plays a sound when GitHub Copilot CLI is not in focus when enabled in the GitHub Copilot CLI settings.
  • Disabled MCP Server – Improves GitHub Copilot CLI startup performance by disabling MCP servers that are not needed for WPA investigations. Since WPA loads Copilot CLI with your configured MCP servers, disabling unused servers reduces startup time and the noise in the context.
  • Screen Reader Optimization – This setting is to provide parity with screen reader optimization settings in GitHub Copilot CLI.

See also

  • Last updated on