IntelliJ developer ships esports analytics tool using JetBrains AI autonomous coding
A developer has shipped an esports performance analysis tool using JetBrains' Junie AI assistant, demonstrating the practical capabilities of autonomous coding agents in production workflows.
The project, ScrIM Coach, analyzes League of Legends match data via the GRID Esports API to generate coaching insights. Built during JetBrains' Sky Limit hackathon, it represents a test case for AI-assisted development within native IDE environments rather than VS Code extensions.
What worked
The developer, a self-described backend engineer, reported success with "Brave Mode," Junie's autonomous operation setting. The assistant handled multi-file refactoring tasks, restructuring 15 files simultaneously while maintaining test coverage and UI functionality.
Notably, the tool generated pixel-perfect frontend implementations from screenshots—a capability gap for developers more comfortable with backend architecture. The assistant also managed git operations including commit message generation and squashing multiple commits.
The developer refactored an initial MVP to follow SOLID principles and clean architecture patterns in a single prompt session. According to the account, functionality remained intact post-refactoring.
The context
Junie scores 53.6% on SWEBench Verified's 500-task benchmark, positioning it competitively against GitHub Copilot and Cursor for autonomous coding tasks. JetBrains integrated Junie into its AI Chat interface in December 2025, consolidating previously separate tools.
The assistant runs on Claude Sonnet 3.7/4 models and maintains project context within JetBrains IDEs including IntelliJ IDEA, PyCharm, and WebStorm. JetBrains made core AI features, including Junie and unlimited code completion, free in 2025.
What's missing
The developer noted the absence of Model Context Protocol (MCP) server support for automated deployment to production environments. Speed remains a documented tradeoff: autonomous task execution takes longer than line-by-line completion tools, though users accept this for complex operations.
The question for enterprise teams: does native IDE integration justify potential speed tradeoffs compared to standalone tools? For organizations already standardized on JetBrains tooling, eliminating context-switching may outweigh raw completion speed.
ScrIM Coach is available under MIT license. The project demonstrates that autonomous AI assistants can handle full-stack development tasks when given sufficient context and model capability.