Towards self-driving codebases · Cursor

We're making a part of our multi-agent research harness available to try today in preview.

Blog / research We're excited by the reaction to our research on scaling long-running autonomous coding . This work started as internal research to push the limits of the current models. As part of the research, we created a new agent harness to orchestrate many thousands of agents and observe their behavior. By last month, our system was stable enough to run continuously for one week, making the vast majority of the commits to our research project (a web browser). This browser was not intended to be used externally and we expected the code to have imperfections. However, even with quirks, the

saved by

• Amir
• Rajan Agarwal
• Ratan Kaliani
• Proby Shandilya
• Rishi Kothari
• Florent Tavernier

related reading

• Scaling long-running autonomous coding · Cursorcursor.com
• Agent swarms and the new model economics · Cursorcursor.com
• Effective harnesses for long-running agents \ Anthropicanthropic.com
• Harness Engineering for Self-Improvement | Lil'Loglilianweng.github.io
• How we built our multi-agent research system \ Anthropicanthropic.com
• Notes on the Software Factorybenedict.dev
• Shipping at Inference-Speed | Peter Steinbergersteipete.me
• Don’t Build Multi-Agents | Cognitioncognition.ai
• Code was our medium for thoughtwattenberger.com
• How to Harness Coding Agents with the Right Infrastructure | Blogalexlavaee.me
• Thoughts on slowing the fuck downmariozechner.at
• Patterns and problems in multiagent systemsanthropic.com