GitHub Copilot Logs
GitHub Copilot writes rolling diagnostic and telemetry logs under ~/.copilot/logs on Windows so its local agent, authentication flow, chat features, and editor integrations can record requests, errors, and extension state changes for troubleshooting. These files typically include plain-text session logs, renderer and agent diagnostics, network error traces, and crash-related output that accumulate over time but are not required for normal operation once they are older. Kudu removes only stale GitHub Copilot log files older than one week, while leaving chat sessions, edit sessions, credentials, and IDE data untouched.
Why clean GitHub Copilot Logs?
- Verbose diagnostic logging during repeated sign-in retries or extension errors can leave hundreds of old text logs behind, showing up as unexpected disk usage in the user profile
- Crash loops or failed background agent restarts often generate the same stack traces over and over, so the logs folder grows quickly even though Copilot itself is no longer using those old files
- When troubleshooting is finished, week-old request and error traces no longer help the extension run better but still clutter searches and backup jobs with thousands of small files
- Large piles of historical logs make it harder to find the current failure in a support zip or manual inspection, because the useful recent entries are buried under stale sessions
- Heavy log churn from chat, inline completions, and network retries can increase file indexing and antivirus scanning activity, which users may notice as brief slowdowns when opening their profile folders
- Old diagnostics can preserve obsolete error patterns from previous extension builds, leading users reviewing the folder to chase problems that no longer exist after an update
Cache paths Kudu targets
Windows
~/.copilot/logs |
Common questions about GitHub Copilot Logs
Download Kudu and reclaim your disk space.
Available on Windows, macOS, and Linux. No account required, no feature gates, no telemetry without consent. All cleaning targets are open source and community-auditable.