Coping with Copilot: Addressing Critical Stability Issues for Engineering Quality Software

Introduction to Copilot's Challenges

GitHub Copilot Pro+, a powerful AI coding assistant, is designed to boost developer productivity. However, recent community discussions highlight significant stability and reliability issues that are hindering rather than helping the development of engineering quality software. Users are reporting a range of critical bugs, from persistent model selection problems to data loss and frustrating rate limits, making it difficult to rely on the service for daily work.

Key Stability and Reliability Issues Reported

A user, 'kryre', initiated a discussion detailing their experience, which paints a picture of an unstable service. The primary concerns include:

• Unreliable Model Switching: Despite explicit selection, Copilot often reverts to default models (e.g., Opus 4.6 x30 fast) within the same agent session. This not only leads to unintended model usage but also consumes quota rapidly without the user's intent.
• Session Persistence Failures: A major concern is the complete loss of chat and agent sessions upon closing and reopening projects. This results in lost work with no recovery options.
• Frequent Request Errors: Developers frequently encounter a 413 Request Entity Too Large error, indicating issues with handling larger contexts.
• Unjustified Rate Limiting: Users are often rate-limited for no clear reason, especially after retrying failed requests. This can lead to complete service blocks, as illustrated by messages like:

  Chat took too long to get ready. Please ensure you are signed in to GitHub and that the extension GitHub.copilot-chat is installed and enabled. Click restart to try again if this issue persists.

• Agent Initialization Failures: Errors such as No activated agent with id 'github.copilot.editsAgent' and Language models loaded: false (rate limit exceeded) indicate problems with the underlying AI agents.
• Application Crashes: The application itself sometimes crashes, further disrupting workflow.

These issues collectively create an environment where Copilot can completely block development work, undermining its promise of enhancing developer productivity and the creation of engineering quality software.

Community-Sourced Workarounds and Solutions

While awaiting official fixes, the community has identified some temporary workarounds:

• Hiding Problematic Models: For persistent model switching issues, 'roshhellwett' suggests going into settings and completely hiding the unwanted model. This forces Copilot to respect other selections.
• Managing Context for 413 Errors: The 413 Request Entity Too Large error is often linked to overly long conversation contexts. Splitting conversations into new chat sessions or reducing the amount of code/text in a single message can help.
• Handling Rate Limits: If rate-limited after retrying failed requests (which count against quota), the only reliable solution is to wait it out (typically 10-30 minutes). Restarting VS Code and re-authenticating the GitHub extension might sometimes clear the "Chat took too long to get ready" error faster.
• Reporting Data Loss: For critical issues like disappearing sessions and lost work, it's highly recommended to report directly to GitHub Support, providing account details and timestamps.

Towards More Reliable Engineering Quality Software

The feedback from the GitHub community underscores the need for greater stability and reliability in AI-powered developer tools. While Copilot offers immense potential, these foundational issues can severely impede developer workflow and the commitment to delivering engineering quality software. As the platform evolves, addressing these core bugs will be crucial for Copilot to truly become an indispensable and trustworthy assistant for developers worldwide.