Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
Kimi K2.7 Code delivers substantial benchmark improvements over its predecessor while cutting reasoning token usage by 30%, making a capable open-weights coding model more efficient and freely accessible.
The benchmark reveals that frontier AI models — including those augmented with Code Agents — effectively fail at large-scale game project engineering, with runtime pass rates collapsing to 5.7%, exposing architectural design as an unsolved bottleneck that compilation-focused improvements cannot address.
The system replaces unconstrained LLM escalation with a structured, forecast-grounded pipeline and introduces a regulator-aligned evaluation metric for false interventions — two gaps the authors identify as absent from existing DeFi supervision approaches.
The paper surfaces a pre-PR coordination layer that existing PR-history analysis cannot see, and provides a concrete substrate and mining toolkit that reduce redundant multi-agent work from 78% to 0% — directly addressing why autonomous agents' PRs are accepted less often despite being produced faster.
By forcing LLM agents to commit their security assumptions as falsifiable assertions and immediately stress-testing them with a fuzzer, Code-Augur replaces opaque agent reasoning with a verifiable, self-correcting audit loop — directly addressing the missed-vulnerability risk the paper identifies as the central weakness of current agentic security analysis.
DSG demonstrates that externalizing search grounding into a shared, MCP-compatible layer can reduce production search costs by over 98% while preserving accuracy, replacing a fixed, opaque model feature with a tunable, provider-agnostic interface.
DIA replaces the multi-party, lossy handoff workflow of enterprise data integration with a fully autonomous, execution-grounded agent system that generalizes across SQL dialects and task categories without task-specific engineering.
The near-universal adoption of tool descriptions contrasts sharply with the low rate of output schemas, revealing a gap in MCP server metadata that affects how reliably agents can interpret and act on tool results.
ProfiLLM demonstrates that an agentic LLM pipeline can move beyond structured numerical features in a live, millisecond-latency industrial dispatcher and produce measurable improvements in real-world GMV and completion rates — validated by a 14-day online A/B test on DiDi's production system.
The work demonstrates that agentic, multi-agent prompt optimization can compound noisy real-world A/B test cycles into statistically robust improvements, offering a practical alternative to gradient-based prompt tuning for open-ended task-oriented dialogue systems.