New Method Helps AI Coding Agents Remember and Follow User Corrections Over Time
Researchers introduced TRACE, a system that converts user corrections into enforced rules for AI coding agents, addressing the problem that agents often repeat the same mistakes across different sessions. Current memory systems like Mem0 fail to enforce user preferences 57.5% of the time, while TRACE reduces preference violations to as low as 2% on new tasks. This matters because AI agents are increasingly used in daily work, and the ability to learn from corrections could significantly improve their reliability and user experience.
A new research paper from arXiv describes TRACE (Test-time Rule Acquisition and Compiled Enforcement), a system designed to help AI coding agents better incorporate user feedback over time. The researchers identified a key problem: while AI agents can access user corrections in memory systems, they frequently violate the same preferences in future sessions. Testing with existing memory solutions like Mem0 showed that 57.5% of applicable user preferences were still violated. TRACE addresses this by automatically converting user corrections into atomic rules and compiling them into runtime checks that must pass before the agent completes tasks. In experiments on coding tasks, TRACE reduced preference violations from 100% to 37.6% on familiar tasks and to just 2% on new, unfamiliar tasks. The researchers made their code and deployable skill publicly available, suggesting the approach could be integrated into existing AI agent systems.
What's missing
The paper does not discuss potential limitations of the TRACE approach, such as how it handles conflicting user corrections, whether rule compilation scales to large numbers of corrections, or how the system performs with non-coding tasks. Additionally, the real-world deployment challenges and user study validation beyond simulated experiments are not addressed.
What different sources said
- arXiv cs.CLCenter
Getting Better at Working With You: Compiling User Corrections into Runtime Enforcement for Coding Agents
Related
Topology-Aware Thermodynamics Improves DNA Probe Specificity Design
Researchers developed a new framework for designing DNA probes that accounts for the spatial organization of matched sequences, not just overall thermodynamic stability. Traditional methods rely on scalar measures like melting temperature and free energy, which miss how mismatches are distributed along the probe. The approach could improve diagnostic accuracy in applications like HPV detection and gene expression profiling.
Study Identifies Optimal Thermal Dose for Combining Focused Ultrasound with Immunotherapy in Tumors
Researchers used multimodal PET imaging to identify an optimal thermal dose range for focused ultrasound ablation that destroys tumor tissue while preserving conditions for immunotherapy delivery. The study found that excessive heating collapses blood vessels needed for antibody access, while insufficient heating fails to adequately reduce tumor burden. The findings could guide clinical design of combination treatments pairing thermal ablation with immunotherapies.
Plant MSH1 Protein Functions as Mismatch-Directed Nuclease for Organelle Genome Maintenance
Researchers have identified the precise mechanism by which the AtMSH1 protein in Arabidopsis plants recognizes and cleaves DNA mismatches and lesions, preventing mutations in organellar genomes. The protein combines a DNA mismatch recognition module with a nuclease domain that makes staggered cuts at specific positions relative to DNA damage. This discovery explains how plants maintain unusually low mutation rates in their mitochondrial and chloroplast DNA compared to other eukaryotes.