Serious UX Issue: Unintended Autonomous PR Creation and UI Lockdown in GitHub Copilot Chat (Plan Mode Violation)

[ogwata]

🏷️ Discussion Type

Product Feedback

💬 Feature/Topic Area

Copilot in GitHub

Body

Description

I'm sharing a critical UX issue where Copilot autonomously created a PR without my consent. Has anyone else experienced this? It’s a significant productivity killer.

I am reporting a critical issue regarding the autonomous behavior of GitHub Copilot (v0.42.3) in VS Code (v1.114.0). Under a "Copilot for Business" seat, the AI agent created a Pull Request (#1) on my repository without my explicit consent or review, causing significant disruption to my workflow.

Issues Observed

1. Unauthorized PR Creation: Despite not clicking any "Apply" or "Create PR" buttons, the agent autonomously created a Pull Request (Welcome to GitHub product feedback #1) on my repository.
2. UI Lockdown: During this process, the model selection dropdown disappeared, and the UI was forced into a "Task" mode, preventing me from switching providers or models.
3. Intent Misalignment: Although I intended to use "GPT-5.4", the system unilaterally used "Claude 3.7 (claude-opus-4.6)" via the agent proxy.
4. Labor Imposition: I had to spend valuable time manually closing the unauthorized PR and cleaning up the unwanted branches.
5. Explicit Mode Violation: Although I had explicitly selected 'Plan Mode' to review the strategy before any changes, the agent bypassed the planning phase and proceeded to 'Act' (implementation and PR creation) without my confirmation.

Supporting Evidence (Attached)

• GitHub Copilot Chat_log.txt: See timestamps around 2026-04-08 17:12:12 showing "Pull request 1 is not visible yet, retrying..." and the use of Claude Copilot Proxy.
• スクリーンショット 2026-04-09 18.50.46.png: Shows the UI locked in "Task" mode with 1060 lines of unintended changes and the "Checkout" button, while the PR had already been created in the background.
• feat(viewer)_ Add AI proofreading panel with open model support and f_ - github.com.png: Proof of the PR (Welcome to GitHub product feedback #1) being created on GitHub autonomously.
• スクリーンショット 2026-04-09 19.12.11.png: Confirmation of the "Task" session being active in the output channel list.
• Claude以外のモデルに切り替えるにはどうすればいい？.txt: Log of my failed attempts to regain control over the UI and switch providers.

Conclusion

This "autonomous" behavior without a human-in-the-loop confirmation is extremely dangerous and violates the developer's control over their codebase. This took away hours of my productivity. I expect a formal investigation into why this autonomous action occurred despite no user confirmation.

Claude以外のモデルに切り替えるにはどうすればいい？.txt

GitHub Copilot Chat_log.txt

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[ogwata]

Additional Evidence Report

During the investigation of this issue, I discovered that the autonomous agent made unauthorized, critical changes to my local environment and caused extreme system resource exhaustion.

1. Unauthorized modification of chatLanguageModels.json

The configuration file defining available AI models in VS Code's global storage was unilaterally overwritten by the agent.

• Exclusion of other models: As shown in the attached file, all metadata for other models (such as Gemini, Grok, and GPT-4o) that should have been listed was wiped out. The file was replaced with a configuration exclusively for the Claude models used by the agent.
• Disruption of other tools: This unauthorized override not only removed model selection from the UI but also caused other extensions, such as Gemini Code Assist, to fail with "LSP client start-up errors."

2. Extreme resource exhaustion and unauthorized process execution

While the agent was active, my hardware resources (Mac with Apple M4) were consumed at an uncontrollable level.

• Critical Swap usage: As shown in the attached Activity Monitor screenshot, the system reached a staggering 56.32 GB of swap usage, pushing the entire OS to the brink of a total hang.
• Autonomous Docker execution: Although I did not launch Docker myself, a massive Docker process was started autonomously in the background to execute the "Task," consuming a vast amount of memory without any user notification or consent.

Conclusion

These findings prove that the agent, in its pursuit of "task completion," prioritized its own execution over the integrity of the user's development environment and hardware stability. This is not merely a functional bug but a significant "overreach" that compromises the safety of the development platform. I strongly request a rigorous investigation into this behavior and the provision of measures to prevent such unauthorized environment modifications.

chatLanguageModels.json.txt