HN dgr8akki 5/8/2026

I lost track of which Claude Code tab needed me, so I built this

Zsh plugin for iTerm2 that monitors Claude Code sessions and highlights tabs needing user attention via color coding.

HN Topfi 5/8/2026

AI Coding plan comparisons based on actual usage

Comparison of AI coding plan pricing and subsidy levels across models, analyzing cost-effectiveness of frontier vs open-weight models for coding tasks.

HN oldnetguy 5/8/2026

AI Is Starting to Build Better AI IEEE Spectrum

Analysis of recursive self-improvement in AI systems, examining how machines can improve AI model development with human oversight.

HN steveharing1 5/8/2026

OpenKB: A Vectorless Knowledge Base for Long Documents

OpenKB is an open-source system using LLMs to compile documents into structured wiki-style knowledge bases with vectorless retrieval for long documents.

HN phoughton 5/8/2026

Vimes Boots and why the right AI evals could save your project

Discussion of evaluation metrics for AI agents and LLMs, using cost-benefit analogy to explain why proper evals prevent expensive failures.

HN rswcf 5/8/2026

The Margin Sandwich: Where the AI Services Dollar Lands

Analysis of how AI is disrupting professional services pricing, examining economic impact on legal and consulting firms.

HN svessi 5/8/2026

Show HN: Claude Control – macOS dashboard for managing Claude Code sessions

Claude Control is a macOS dashboard for monitoring and managing multiple Claude Code sessions with live status and git tracking.

HN tzury 5/8/2026

CUDA-oxide an experimental Rust-to-CUDA compiler

cuda-oxide is an experimental Rust compiler backend enabling GPU kernel compilation in pure Rust without DSLs or bindings.

HN Mapika 5/8/2026

Show HN: A Local-First Agentic Knowledge Manager

Kept is a local-first desktop app that archives AI conversations as Markdown files with search, browsing, and graph visualization. Supports ChatGPT, Claude, Gemini, and other LLMs.

HN samagragune 5/8/2026

Show HN: The agent which teaches you while you build

Contral is an AI agent and IDE that teaches developers while they code through real-time explanations and interactive debugging.

HN franze 5/8/2026

Show HN: Airplane AI – Local NDA Safe AI Powered by Gemma

Airplane AI is a local-only LLM application powered by Gemma that runs offline on-device without cloud dependencies or accounts.

HN damianofalcioni 5/8/2026

Pi-for-Word: Pi Agent as Add-In for M365 Office Word

Pi4Word integrates Pi Agent-powered AI assistant into Microsoft Word as task pane add-in, leveraging @mariozechner/pi-agent-core with tool support and streaming capabilities.

HN eamag 5/8/2026

Notes from Inside China AI Labs

Personal observations from visiting Chinese AI research labs. Anecdotal insights into AI ecosystem and research culture.

HN idradev 5/8/2026

Diagrammer: Tell the model, get the diagram or the graphical map

Open-source, local-first viewer for LLM-generated diagrams (mindmaps, flowcharts, ER diagrams, etc.). Schema-driven rendering with no cloud dependency or vendor lock-in.

HN onder_ceylan 5/8/2026

The complete Claude Code course for engineers and technical founders

Course on using Claude Code with focus on building TDD pipelines, guardrails, and verification systems for production-grade AI-generated code.

HN matt_callmann 5/8/2026

Show HN: Runs AI coding agents inside isolated Docker containers

AI coding agent (pi) running in isolated Docker containers with no root access or privilege escalation, designed for safe local execution.

HN kkarpkkarp 5/8/2026

Show HN: I instructed AI to create tool to clean AI-generated text: undsh.com

Tool using Cursor AI to clean formatting artifacts from AI-generated text in documents and code.

HN nlpnerd 5/8/2026

AI-Native Hedge Funds Are Possible and Profitable Just Not the Next Unicorn

Analysis of AI adoption in hedge funds and quantitative trading strategies. Tangential to core AI/ML interests.

HN mike-cardwell 5/8/2026

Auth Proxy Injection for LLMs

Python sandbox for running Claude LLM with proxy injection and configuration options for containerized execution.

HN anujbans 5/8/2026

SubQ: Sub-quadratic LLM built for 12M-token reasoning

SubQ: long-context LLM API supporting 12M tokens at linear cost for code agents and repository analysis.

HN geox 5/8/2026

Why does AI like goblins and Japan so much?

Analysis of ChatGPT bias toward goblin and Japan references, documented across model versions.

HN lucid-dev 5/8/2026

Show HN: When the LLM Accidentally

User observation of LLM output behavior showing internal reasoning exposed in responses. Informal anecdote without technical depth.

HN mfarias 5/8/2026

BotScript – a TypeScript superset for code mostly written by bots

TypeScript superset language designed for code generated by AI/LLMs, with browser-based playground and compiler.

Ax Alex Oesterling, Donghao Ren, Yannick Assogba, Dominik Moritz, Sunnie S. Y. Kim, Leon Gatys, Fred Hohman 5/8/2026

Understanding Annotator Safety Policy with Interpretability

Analysis of annotation disagreement sources in AI safety policies: operational failures, policy ambiguity, and value pluralism.

Ax Robert Washbourne, Rishi Iyer, Tomas Figliolia, Henry Zheng, Ryan Lorig-Roach, Sungyeon Yang, Pritish Yuvraj, Quentin Anthony, Yury Tokpanov, Xiao Yang, Ganesh Nanduru, Stephen Ebert, Praneeth Medepalli, Skyler Szot, Srivatsan Rajagopal, Alex Ong, Bhavana Mehta, Beren Millidge 5/8/2026

ZAYA1-8B Technical Report

ZAYA1-8B is a reasoning-focused mixture-of-experts model with 8B total parameters built on AMD compute, matching larger models on mathematics and coding benchmarks.

Ax Krti Tallam 5/8/2026

Partial Evidence Bench: Benchmarking Authorization-Limited Evidence in Agentic Systems

Partial Evidence Bench introduces a deterministic benchmark for measuring failures in agentic systems operating under access control and authorization constraints.

Ax Aymen Echarghaoui, Dongxia Wu, Emily B. Fox 5/8/2026

BALAR : A Bayesian Agentic Loop for Active Reasoning

BALAR proposes a Bayesian agentic loop algorithm for active reasoning in LLMs, enabling principled information gathering and question generation in multi-turn interactions without fine-tuning.

Ax Jiechen Li, Catherine A. Barry, Rishika Randev, Janet Chen, Ella Jorgensen, Brinnae Bent 5/8/2026

When Helpfulness Becomes Sycophancy: Sycophancy is a Boundary Failure Between Social Alignment and Epistemic Integrity in Large Language Models

Position paper on sycophancy as boundary failure between social alignment and epistemic integrity in LLMs, broader than surface-level agreement behavior.

Ax Mohamed Salim Aissi, Clemence Grislain, Clement Romac, Laure Soulier, Mohamed Chetouani, Olivier Sigaud, Nicolas Thome 5/8/2026

PRISM: Perception Reasoning Interleaved for Sequential Decision Making

Embodied agent framework coupling vision-language and language models via dynamic QA pipeline for multimodal sequential decision making.

Ax Yang Shu, Yingmin Liu, Zequn Xie 5/8/2026

Agentic Retrieval-Augmented Generation for Financial Document Question Answering

Agentic RAG framework for financial QA combining multi-step reasoning over tables, text, and footnotes with dynamic retrieval and decision making.

Ax Jesse A. Rodr\'iguez 5/8/2026

LaTA: A Drop-in, FERPA-Compliant Local-LLM Autograder for Upper-Division STEM Coursework

Open-source on-premises LLM autograder for STEM courses running on commodity hardware, FERPA-compliant alternative to cloud-based grading APIs.

Ax Haoyang Xie, Xinyuan Wang, Yancheng Wang, Puda Zhao, Feng Ju 5/8/2026

From History to State: Constant-Context Skill Learning for LLM Agents

LLM agent framework maintaining constant context size by converting long interaction histories to compressed skill state, balancing privacy and capability.

Ax Alif Al Hasan 5/8/2026

The Geopolitics of AI Safety: A Causal Analysis of Regional LLM Bias

Causal analysis of regional LLM bias using probabilistic graphical models to audit safety mechanisms beyond observational fairness metrics.

Ax Krti Tallam 5/8/2026

Authorization Propagation in Multi-Agent AI Systems: Identity Governance as Infrastructure

Authorization propagation framework for multi-agent AI systems addressing identity governance and access control across task delegation and data boundaries.

Ax Titouan Duston, Jiashu Liang, Yuanheng Wang, Weihao Gao, Xuelan Wen, Nan Sheng, Weiluo Ren, Yang Sun, Yixiao Chen 5/8/2026

Agentic Discovery of Exchange-Correlation Density Functionals

Agentic search system using LLMs to systematically discover exchange-correlation density functionals, automating human-driven DFT design loops.

Ax Allessia Chiappetta, Robert Mahari 5/8/2026

Intentionality is a Design Decision: Measuring Functional Intentionality for Accountable AI Systems

Position paper defining intentionality as behavioral profile for accountable AI systems with criteria: purpose, foresight, volition, temporal commitment, coherence.

Ax Mahyar Alinejad, Yue Wang, Amrit Singh Bedi, George Atia 5/8/2026

LANTERN: LLM-Augmented Neurosymbolic Transfer with Experience-Gated Reasoning Networks

Multi-source neurosymbolic transfer learning framework combining LLMs and reinforcement learning with adaptive knowledge integration mechanisms.

Ax Denys Katerenchuk, Pablo Duboue, Keelan Evanini, David Gondek, Nithin Govindugari, Olivier Allauzen, Joshua Baptiste, David J More, Joshua Schechter 5/8/2026

FinRAG-12B: A Production-Validated Recipe for Grounded Question Answering in Banking

Production-validated grounded QA framework for banking domain optimizing LLM accuracy, citation grounding, and calibrated refusal under regulatory constraints.

Ax Woojin Lee, Pranav Mekkoth, Ye Tian, Onat Gungor, Tajana Rosing 5/8/2026

FoodCHA: Multi-Modal LLM Agent for Fine-Grained Food Analysis

Multi-modal LLM agent for fine-grained food recognition and analysis handling intra-class similarity and multiple items per image.

Ax Susheel Suresh, Hazel Mak, Shangpo Chou, Fred Kroon, Sahil Bhatnagar 5/8/2026

AgenticRAG: Agentic Retrieval for Enterprise Knowledge Bases

Agentic RAG system for enterprise knowledge base retrieval and analysis that reduces search bottlenecks via LLM-controlled search policies.

Ax Hyobin Park, Taeseop Kim, Dong-Geol Choi 5/8/2026

SPARK: Self-Play with Asymmetric Reward from Knowledge Graphs

Self-play reinforcement learning framework using knowledge graphs for asymmetric rewards to extend RL to scientific literature analysis.

Ax Siqi Zhu 5/8/2026

Who Prices Cognitive Labor in the Age of Agents? A Position on Compute-Anchored Wages

Position paper on AI agent economics arguing agents are production technology, not labor, with implications for wage theory and policy.

Ax Sai Babu Patarlapalli, Surya Teja Avvaru 5/8/2026

BitCal-TTS: Bit-Calibrated Test-Time Scaling for Quantized Reasoning Models

Post-training quantization technique for reasoning models with calibrated test-time scaling to maintain inference quality under memory/latency constraints.

Ax Langlin Huang, Chengsong Huang, Jinyuan Li, Donghong Cai, Yuyi Yang, Jiaxin Huang 5/8/2026

Nonsense Helps: Prompt Space Perturbation Broadens Reasoning Exploration

LLM reasoning improvement via prompt perturbation to address zero-advantage problem in reinforcement learning with verifiable rewards like GRPO.

Ax Chuan-Xian Ren, Cheng-Jun Guo, Hong Yan 5/8/2026

Locality-aware Private Class Identification for Domain Adaptation with Extreme Label Shift

Domain adaptation approach for handling label shift and private classes in transfer learning scenarios with non-overlapping class spaces.

Ax Yishuo Yuan, Jiayi Sheng, Sirui Zeng, Jiaqi Wang, Jiaheng Liu 5/8/2026

AlphaCrafter: A Full-Stack Multi-Agent Framework for Cross-Sectional Quantitative Trading

Multi-agent framework for quantitative trading coupling factor discovery, regime-adaptive selection, and risk-constrained execution.

Ax Junfeng Liao, Qizhou Wang, Jianing Zhu, Bo Du, Rui Yan, Xiuying Chen 5/8/2026

Belief Memory: Agent Memory Under Partial Observability

Memory architecture for LLM agents handling partial observability via belief distributions instead of deterministic conclusions.

Ax Zehao Deng, Tianjie Ju, Zheng Wu, Liangbo He, Jun Lan, Huijia Zhu, Weiqiang Wang, Zhuosheng Zhang 5/8/2026

Causal Probing for Internal Visual Representations in Multimodal Large Language Models

Causal probing framework using activation steering to study internal visual representations in MLLMs.

Ax Ran Bi, Shiyao Wei, Yuanyiyi Zhou 5/8/2026

Prober.ai: Gated Inquiry-Based Feedback via LLM-Constrained Personas for Argumentative Writing Development

Web-based writing environment using LLM-constrained personas for argumentative feedback development.

Ax Jiarui Zhong, Hong Cai Chen 5/8/2026

Text-Graph Synergy: A Bidirectional Verification and Completion Framework for RAG

Bidirectional text-graph verification framework for RAG improving factual grounding and multi-hop reasoning in LLMs.