CodeScene

A legacy uplift that would traditionally have been a 12–18-month, high-risk engineering project was completed in three weeks of part-time work for roughly $4,000 in AI costs. The entire codebase was refactored to eliminate technical debt, bringing the code to a level where new features could be added safely with AI. Read the full case study: https://lnkd.in/eUN7rB6H #aicoding #technicaldebt #codequality #agenticai #aiagents

CodeScene is involved in several research projects! 🚀 One of them is NextG2Com, a competence center led by Lund University and co-funded by Vinnova. Its partners range from Ericsson, Sony, Volvo Cars, Bosch and Tetra Pak to startups such as Cognibotics, Acconeer and remotiveLabs. It's a nice mix of skill sets, and the discussions are always inspiring. In the middle of the agentic revolution, it's valuable to have forums where we can meet and openly discuss what it means in practice. Last week we joined the center's annual meeting in Lund. We brought two things this year: - We demoed our refactored Street Fighter version for the first time. AI agents guided by our CodeHealth MCP server removed all identified code smells from the entire codebase in three weeks, taking CodeHealth from 5.6 to 10.0. Visitors could play the refactored game right at our booth! - Nadim Hagatulah presented his latest work on how CodeHealth shapes the outcomes of coding agents, joint work with Emma Söderberg and Adam Tornhill. The results will be submitted to a journal this week!

Claude har precis fått stöd för Mods, där man med typescript enkelt kan anpassa gränssnittet både i desktop-appen och cli. Passade på att testa att göra en mod som hela tiden visar code health score för diffen som claude jobbar med, samt en sidopanel där man kan se detaljer fil för fil. CodeScene

12–18 months. That's how long refactoring a 300K-line codebase traditionally takes, and it's a high-risk project. Adam Tornhill Markus Borg and Daniel Webb did it for Street Fighter III codebase in 3 weeks, with AI. Code Health went from 5.6 to a perfect 10.0. The code is now AI-ready, with ~70% fewer AI-induced defects and ~45% less token waste on future features. Total cost? about $4,000 in tokens. 🙌 Where to watch the full episode: • Substack: https://lnkd.in/eEDyauDj • YouTube: https://lnkd.in/eU27n_mg • Spotify: https://lnkd.in/dDDC_KSb

📣 loveAI speaker announcement! 📣   AI can help teams write code faster, but what happens when the code it builds on is already slowing teams down?   At loveAI, Adam Tornhill, CTO and founder of CodeScene, will explore how code health shapes the value organisations can get from AI coding.   Drawing on large-scale research, Adam will share practical ways to make agentic coding work across enterprise teams.   AI Coding in the Enterprise: Faster Coding Requires Better Code   Join us at loveAI: https://lnkd.in/e26M8fjw #ai #techgrowth #aiconference

A legacy uplift that would traditionally have been a 12–18-month, high-risk engineering project was completed in three weeks of part-time work for roughly $4,000 in AI costs. The entire codebase was refactored to eliminate technical debt, bringing the code to a level where new features could be added safely with AI. The CodeHealth MCP Server served as the objective quality signal and the agentic feedback loop. Read the full case study: https://lnkd.in/eUN7rB6H Kudos to Adam Tornhill, Markus Borg and Daniel Webb #aiagents #agenticai #technicaldebt #codequality #aicoding

This is my biggest AI wow moment to date: 300K lines of complex code refactored in three weeks to perfect Code Health. In the past, uplifting a legacy codebase would have been an expensive, high-risk project needing 12–18 months with a team of experts. Now we did it in a fraction of that time for just $4,000 worth of tokens. Some of the highlights from this case study: * The case study demonstrates agentic refactoring at scale on a real-world, non-trivial codebase: Street Fighter III: 3rd Strike. * We refactored the whole codebase to eradicate all application code technical debt. * The code was brought to a level where new features could be added safely with AI. * Functional correctness was verified via a replay-trace harness. (And yes: we could still play the game as before.) * The CodeHealth MCP Server was used as the objective quality signal and agentic feedback loop. But perhaps the most exciting part is the process we used. We had agents iteratively building up a refactoring playbook. That way, agentic refactoring tasks become progressively more effective. As part of that, we discovered novel codebase-specific refactoring rules courtesy of the AI. These were remarkably useful, yet structurally different from the code transformations an expert human would consider. I'll cover all of that in my new article: https://lnkd.in/eBgaX-ET Big thanks to the CodeScene research team, Markus Borg, and Daniel Webb for breaking this new ground.