Fetching from the wire…
01
02
03
04
05
06
07
08
Agents2026-09-22 · source-backed
LazyAgent replaces plan-then-execute-when-ready with a live goal-derived demanded set: a backward closure from requested outputs refreshed as state changes, so a ready node materializes only when the active goal needs it, turning repeated local judgments into one linear-time graph analysis plus constant-time membership tests. Adding one unrelated product raised the eager baseline's bill 22.5% and LazyAgent's by zero. It saves 42.0% of measured CPU on production scientific workflows and 51.7% of container time on a live release gate spanning four repositories. (arXiv 2609.23058)
Each link below shares sources, entities, or timing with this story.
arXiv 2609.09769 argues that relying on the static issue description biases reasoning toward the narrow scope of that text. Adding dynamic behavioral analysis gets 72.8% function-localization accuracy while staying cost-competitive, and resolves 7 issues the top baselines miss...
arXiv 2607.07368 formalizes distributed attacks where multiple agents jointly pursue a malicious goal, evaluated on FakeLab (9 services, 86 benign tasks, 4 attack objectives). Adding coordinating agents makes per-agent monitoring *less* likely to catch any individual one, and...
Created August 31, MIT, 850 stars with 184 forks, a 22% fork-to-star ratio that suggests people are running it rather than bookmarking it. It pairs a model with a deterministic pure-Python RE toolkit (PE/ELF/Mach-O parsing, x86/x64/ARM/ARM64 disassembly, AOB scanning, CPU emul...
A controlled study ran five Qwen models over eight cases against a DWSIM simulator, 120 slots per arm, with one instruction as the only difference: request a fresh simulation after a substantive modification. No hard gate. Re-verification happened in 94 of 120 guided slots aga...
arXiv 2608.26197 stacked finite-state control, forced tool selection, output validation and bounded retries on two open-weight models, and got mixed results across all four model-task cells. Adding structured planning, where the plan is checked against a fixed schema before an...
arXiv 2608.06337 settles an open question on the monotone-adversary model, where an adversary appends examples all labeled correctly by the target hypothesis but chosen after seeing the clean sample. The extra logarithmic factor is inherent, not algorithmic: minimax expected e...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.