We just completed a company-wide rollout of GitHub Copilot to about 200 developers, mostly in VS Code. The overall boost in productivity is undeniable, but we did hit some snags with plugin conflicts that I thought would be valuable to share.
The main issue wasn't Copilot itself, but how it interacted with other language-specific and linting tools. The most common complaint was sudden spikes in memory usage and lagging IntelliSense. After digging into logs and user reports, a pattern emerged. The biggest culprit was having multiple plugins that all tried to provide language server features or auto-completions. For example, teams using Python with Pylance, along with older Python extensions that had their own linting engines, saw CPU usage spike until we disabled the redundant features. Similarly, some JavaScript/TypeScript setups with heavy ESLint plugins and Copilot would cause the editor to freeze during saves.
We also saw conflicts with certain "AI-powered" snippets plugins and even some niche formatting tools. It seems like when two or more plugins are trying to analyze or suggest code at the same trigger point, things can get contentious. Our solution was to audit the recommended plugin list per language and enforce a "one language server" rule where possible, letting Copilot layer on top of that single source of truth.
Has anyone else done a large rollout and seen similar patterns? I'm particularly curious about conflicts in JetBrains IDEs or with Neovim setups, as we had fewer of those in our mix.
✌️
✌️