Isolater - Feed

HN weinzierl 5/7/2026

Symposium makes Rust dependencies actionable for AI agents

Tool enabling AI agents to discover and use Rust crate plugins, skills, and MCP servers with project-specific context.

HN johnymontana 5/7/2026

Create Context Graph: Scaffold AI Agents with Graph Memory

Neo4j Labs CLI creating full-stack context graph agent apps with knowledge graphs, decision traces, and visualization for AI agents.

HN esubaalew 5/7/2026

Show HN: Appctl LLM tools for apps you have

CLI tool generating MCP tools from OpenAI docs or databases, enabling LLMs to control custom applications via terminal or web UI.

HN throawayonthe 5/7/2026

Cursed Browser: Rendering Engine Using Visual-LLMs

Experimental open-source browser rendering engine that uses visual LLMs to interpret HTML instead of traditional rendering. Proof-of-concept.

HN segmenta 5/7/2026

How Unsloth and Nvidia made LLM training 25% faster on consumer GPUs

Unsloth and Nvidia optimization achieving 25% faster LLM training on consumer GPUs.

HN doener 5/7/2026

Stackit and neuland.ai aim to provide LLMs in European data centers

Stackit and neuland.ai partner to provide LLM access via German data centers for GDPR-compliant AI deployment.

HN rcarmo 5/7/2026

Notes on GPT 5.x Model Regressions

Analysis of GPT-5.5 code generation regressions where the model makes unrequested changes and improvements to unrelated code.

HN darkrishabh 5/7/2026

Show HN: Agent-skills-eval – Test whether Agent Skills improve outputs

Agent-skills-eval is a test runner for evaluating whether Agent Skills improve model outputs through empirical testing of SKILL.md files.

HN krautburglar 5/7/2026

Architectural Framework for Agentic AI in Identity and Eligibility

Microsoft architectural framework for deploying agentic AI in public sector identity and eligibility verification systems.

LB github.com via gcv 5/7/2026

The Agent Harness Framework

Flue is a TypeScript framework for building AI agents with a built-in harness, designed as a headless, programmable alternative to Claude Code.

HN eyalitki 5/7/2026

MRC Protocol: Supercomputer networking to accelerate large scale AI training

MRC Protocol is an open networking standard developed by OpenAI, AMD, Broadcom, Intel, Microsoft, and NVIDIA to improve GPU cluster performance for large-scale AI training.

HN andres 5/7/2026

Show HN: Kstack – Skill pack for monitoring/troubleshooting K8s in Claude Code

Kstack is a skill pack for Claude Code enabling Kubernetes monitoring and troubleshooting through reusable commands like /investigate and /audit-security.

HN EGreg 5/7/2026

LAWS: A new transform operation turning LLM inference into cheap cache lookups

LAWS is a research paper on a transform operation that converts LLM inference into efficient cache lookups to reduce computational cost.

HN GregKonush 5/7/2026

Show HN: Bilig, headless spreadsheet engine for Node services and agents

Bilig is a local-first spreadsheet engine for Node services and agents with formula parsing, binary sync protocol, and agent API support.

Ax Cheng Qian, Hyeonjeong Ha, Jiayu Liu, Jeonghwan Kim, Jiateng Liu, Bingxuan Li, Aditi Tiwari, Dwip Dalal, Zhenhailong Wang, Xiusi Chen, Mahdi Namazifar, Yunzhu Li, Heng Ji 5/7/2026

CreativityBench: Evaluating Agent Creative Reasoning via Affordance-Based Tool Repurposing

CreativityBench evaluating LLM creative problem-solving through tool repurposing based on affordance reasoning rather than canonical usage.

Ax Kerri Prinos, Lilianne Brush, Cameron Denton, Zhanqi Wang, Joshua Knox, Snehal Antani, Anton Foltz, Amy Villase\~nor 5/7/2026

Stable Agentic Control: Tool-Mediated LLM Architecture for Autonomous Cyber Defense

Tool-mediated LLM agent architecture with formal guarantees using Stackelberg game theory for autonomous cyber defense in SOCs.

Ax Hao Liu, Xiao-Wen Yang, Atharva Sehgal, Yixin Wang, Lan-Zhe Guo, Yu-Feng Li, Yisong Yue 5/7/2026

Programmatic Context Augmentation for LLM-based Symbolic Regression

Programmatic context augmentation improving LLM-based symbolic regression for mathematical expression discovery in scientific datasets.

Ax Reshabh K Sharma, Gaurav Mittal, Yu Hu 5/7/2026

Learning Correct Behavior from Examples: Validating Sequential Execution in Autonomous Agents

Algorithm learning correct sequential agent behavior from 2-10 execution traces using dominator analysis for validation.

Ax Spandan Garg, Vikram Nitin, Yufan Huang 5/7/2026

Terminus-4B: Can a Smaller Model Replace Frontier LLMs at Agentic Execution Tasks?

Terminus-4B 4B parameter model evaluated as replacement for frontier LLMs in agentic subtasks like code execution and debugging.

Ax Alexandria K. Vail, Marcelo Cicconet, Katie Aafjes-van Doorn, Ryan Maroney, Marc Aafjes 5/7/2026

ADAPTS: Agentic Decomposition for Automated Protocol-agnostic Tracking of Symptoms

ADAPTS mixture-of-agents framework for automated depression and anxiety severity rating from clinical interviews via decomposed reasoning.

Ax Hongkun Yu 5/7/2026

Evaluating Prompting and Execution-Based Methods for Deterministic Computation in LLMs

Evaluation of prompting strategies including CoT and PoT for deterministic computation accuracy in LLMs on mathematical tasks.

Ax Masafumi Oyamada, Kunihiro Takeoka, Kosuke Akimoto, Ryoma Obara, Masafumi Enomoto, Haochen Zhang, Daichi Haraguchi, Takuya Tamura 5/7/2026

cotomi Act: Learning to Automate Work by Watching You

Cotomi Act browser agent learning from user behavior observation with adaptive execution achieving 80.4% task success via multi-step execution.

Ax Zuoyu Zhang, Yancheng Zhu 5/7/2026

Enhancing Agent Safety Judgment: Controlled Benchmark Rewriting and Analogical Reasoning for Deceptive Out-of-Distribution Scenarios

ROME framework for red-teaming tool-using agents with controlled benchmark rewriting to test deceptive safety scenarios.

Ax Bo-Wen Zhang, Jin Ye, Peng-Yu Hua, Jia-Wei Cao, Jie-Jing Shao, Yu-Feng Li, Lan-Zhe Guo 5/7/2026

Revisiting the Travel Planning Capabilities of Large Language Models

Benchmark decomposing travel planning into five sub-capabilities to evaluate LLM long-horizon reasoning and tool use deficits.

Ax Tong Guo, Caishun Chen, Yew Soon Ong 5/7/2026

Automated Large-scale CVRP Solver Design via LLM-assisted Flexible MCTS

LLM-assisted flexible MCTS approach for automated algorithm design to solve large-scale vehicle routing problems with minimal expert input.

Ax Xutao Mao, Jinman Zhao, Gerald Penn, Cong Wang 5/7/2026

What Happens Inside Agent Memory? Circuit Analysis from Emergence to Diagnosis

Circuit analysis of LLM agent memory systems across Qwen models and frameworks, diagnosing write-manage-read pipeline failures.

Ax Jiahao Wang, Mingyue Cheng, Yitong Zhou, Qingyang Mao, Xiaoyu Tao, Qi Liu, Enhong Chen 5/7/2026

GeoDecider: A Coarse-to-Fine Agentic Workflow for Explainable Lithology Classification

GeoDecider agentic workflow applying multi-step reasoning and tool-use to lithology classification in geoscience applications.

Ax Srinath Perera, Kaviru Hapuarachchi, Frank Leymann, Rania Khalaf 5/7/2026

Robust Agent Compensation (RAC): Teaching AI Agents to Compensate

Robust Agent Compensation framework enabling safe agent execution with recovery capabilities via logging, compatible with LangGraph and other agent frameworks.

Ax Shule Lu, Yujing Wang, Hainan Zhang, Xiaoshan Yang, Hongwei Zheng, Yongxin Tong, Changsheng Xu, Zhiming Zheng 5/7/2026

Replacing Parameters with Preferences: Federated Alignment of Heterogeneous Vision-Language Models

Federated learning approach for aligning heterogeneous vision-language models without centralizing data across privacy-sensitive domains.

Ax Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, Tae Yoon Lim, Dongwan Kang, Hwanil Choi, Soonyoung Lee, Wonbin Ahn 5/7/2026

FinSTaR: Towards Financial Reasoning with Time Series Reasoning Models

Taxonomy and methods for time series reasoning models applied to financial domain with multi-entity analysis and future prediction.

Ax Zirui Tang, Xuanhe Zhou, Yumou Liu, Linchun Li, Weizheng Wang, Hongzhang Huang, Jun Zhou, Jiachen Song, Shaoli Yu, Jinqi Wang, Zihang Zhou, Hongyi Zhou, Yuting Lv, Jinyang Li, Jiashuo Liu, Ruoyu Chen, Chunwei Liu, GuoLiang Li, Jihua Kang, Fan Wu 5/7/2026

Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies

Benchmark for evaluating AI agents on workspace tasks requiring reasoning over file dependencies in realistic work environments.

Ax Chenchen Yuan, Zheyu Zhang, Gjergji Kasneci 5/7/2026

Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models

Inference-time steering of LLM moral reasoning via convergent-divergent routing at minimal branch points in transformer blocks.

Ax Robert Gieselmann, Henrike von Huelsen, Mihai Samson, Marie-Christine Meyer, Dariusz Piotrowski, Oleksandr Radomskyi, Justin Okamoto, Turan Gojayev, Michael Painter, Gavin Brown, Federico Pecora, Jeremy L. Wyatt 5/7/2026

Self-Improvement for Fast, High-Quality Plan Generation

Self-improvement approach for generating high-quality plans in sub-exponential time using transformer-based generative models.

Ax Jie Ou, Jinyu Guo, Shiyao Guo, Yuang Li, Ruiqi Wu, Zhaokun Wang, Wenyi Li, Wenhong Tian 5/7/2026

AdapShot: Adaptive Many-Shot In-Context Learning with Semantic-Aware KV Cache Reuse

Adaptive many-shot in-context learning with semantic-aware KV cache reuse for varying query difficulty in LLMs.

Ax Bronislav Sidik, Lior Rokach 5/7/2026

MEMTIER: Tiered Memory Architecture and Retrieval Bottleneck Analysis for Long-Running Autonomous AI Agents

Tiered memory architecture for long-running autonomous agents with structured episodic storage and weighted retrieval to prevent degradation over 72+ hours.

Ax Yiding Ma, Chengyun Ruan, Kaibo Huang, Zhongliang Yang, Linna Zhou 5/7/2026

OracleProto: A Reproducible Framework for Benchmarking LLM Native Forecasting via Knowledge Cutoff and Temporal Masking

Reproducible benchmarking framework for evaluating LLM forecasting capabilities using knowledge cutoff and temporal masking.

Ax Haoxi Li, Qinglin Hou, Jianfei Ma, Jinxiang Lai, Tao Han, Sikai Bai, Jingcai Guo, Jie Zhang, Song Guo 5/7/2026

What You Think is What You See: Driving Exploration in VLM Agents via Visual-Linguistic Curiosity

VLM agents use visual-linguistic curiosity to drive exploration in partially observable environments via mental simulation before acting.

Ax Andrea Iannoli, Lorenzo Gigli, Luca Sciullo, Angelo Trotta, Marco Di Felice 5/7/2026

Say the Mission, Execute the Swarm: Agent-Enhanced LLM Reasoning in the Web-of-Drones

LLM framework for UAV swarm control using natural language mission descriptions with agent-enhanced reasoning for real-time closed-loop execution.

Ax Jiale Chang, Yuxiang Ren 5/7/2026

ScrapMem: A Bio-inspired Framework for On-device Personalized Agent Memory via Optical Forgetting

Bio-inspired memory framework for LLM agents on edge devices using optical forgetting to compress multimodal memories while reducing storage costs.

Ax Chandan Singh, Yan Shuo Tan, Weijia Xu, Zelalem Gero, Weiwei Yang, Michel Galley, Jianfeng Gao 5/7/2026

Agentic-imodels: Evolving agentic interpretability tools via autoresearch

Autoresearch loop evolving interpretability tools specifically designed for agents rather than humans to conduct autonomous data science work.

Ax Munkhdegerekh Batzorig, Purevbaatar Ganbold, Kyungbin Park, Pilkong Jeong, Kangbin 5/7/2026

Mechanical Conscience: A Mathematical Framework for Dependability of Machine Intelligenc

Mathematical framework for ensuring dependability in distributed collaborative intelligence systems where local agent decisions compose into acceptable global behaviors.

Ax Tianyang Han, Hengyu Shi, Junjie Hu, Xu Yang, Zhiling Wang, Junhao Su 5/7/2026

Correct Is Not Enough: Training Reasoning Planners with Executor-Grounded Rewards

Training reasoning planners with executor-grounded rewards beyond correctness to ensure faithful reasoning traces in LLM reasoning chains.

Ax Shuyue Stella Li, Rui Xin, Teng Xiao, Yike Wang, Rulin Shao, Zoey Hao, Melanie Sclar, Sewoong Oh, Faeze Brahman, Pang Wei Koh, Yulia Tsvetkov 5/7/2026

EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative Rubrics

Language models self-improve through co-evolved discriminative rubrics without external supervision, eliminating reliance on human annotations or proprietary APIs.

Ax Pratik Honavar, Tejpratap GVSL 5/7/2026

QKVShare: Quantized KV-Cache Handoff for Multi-Agent On-Device LLMs

Framework for efficient KV-cache handoff between multi-agent LLM systems on edge devices using quantization.

Ax Jie Zhou, Qin Chen, Liang He 5/7/2026

Contextual Multi-Objective Optimization: Rethinking Objectives in Frontier AI Systems

Argues for context-dependent objective optimization in frontier AI systems for open-ended and long-horizon agent tasks.

Ax Kishan Athrey, Ramin Pishehvar, Brian Riordan, Mahesh Viswanathan 5/7/2026

From Intent to Execution: Composing Agentic Workflows with Agent Recommendation

Framework for automated composition of multi-agent systems with agent recommendation replacing manual workflow creation.

Ax Dutao Zhang, Tian Liao 5/7/2026

An Agent-Oriented Pluggable Experience-RAG Skill for Experience-Driven Retrieval Strategy Orchestration

Pluggable agent skill for dynamic retrieval strategy orchestration across heterogeneous tasks in RAG systems.

Ax Joseph Breda, Fadi Yousif, Beszel Hawkins, Marinela Cotoi, Miao Liu, Ray Luo, Po-Hsuan Cameron Chen, Mike Schaekermann, Samuel Schmidgall, Xin Liu, Girish Narayanswamy, Samuel Solomon, Maxwell A. Xu, Xiaoran Fan, Longfei Shangguan, Anran Wang, Bhavna Daryani, Buddy Herkenham, Cara Tan, Mark Malhotra, Shwetak Patel, John B. Hernandez, Quang Duong, Yun Liu, Zach Wasson, Dimitrios Antos, Bob Lou, Matthew Thompson, Jonathan Richina, Anupam Pathak, Nichole Young-Lin, Jake Sunshine, Daniel McDuff 5/7/2026