HN kenshi144 4/20/2026

Show HN: Naptrace – find structural twins of a CVE in your codebase

Naptrace finds structural CVE pattern twins in codebases using code property graphs and LLMs for security analysis.

HN cebert 4/20/2026

Amazon's AI boom is creating mess of duplicate tools and data inside the company

Internal Amazon document reveals AI tool explosion creating software and data duplication problems within the company.

HN giuliomagnifico 4/20/2026

68M AI Crawler Visits Show What Drives AI Search Visibility

Analysis of 68M AI crawler visits across 858k sites reveals patterns in AI search visibility and crawler behavior for SEO optimization.

HN sneefle 4/20/2026

Show HN: AI Applyd – a bot that applies to jobs for you

AI Applyd: autonomous agent that applies to job postings across multiple platforms, optimizes resumes, and bypasses ATS.

HN kkapelon 4/20/2026

Your "AI" Pull Request was rejected and you don't know why

Technical guide explaining why AI-generated pull requests are rejected by open-source maintainers.

HN jamescollinssp 4/20/2026

SerpApi Claude Code Skill

Claude Code plugin integrating SerpApi search capabilities, enabling Claude to query 100+ search engines.

HN popey 4/20/2026

A Proposed Framework for Evaluating AI Agent Skills

Framework for evaluating AI agent skills: reusable instruction bundles and their impact on agent performance.

HN shannon-data-ai 4/20/2026

ShannonBase is database agent platform

ShannonBase: database agent platform. AI agents for database operations and queries.

HN lizzieyo 4/20/2026

The AI-Ready Product Data Framework for B2B Commerce

B2B commerce product data framework for AI with expert insights on data preparation, enrichment, and scalable AI applications.

HN playfultones 4/20/2026

Show HN: Busybee - a FIFO build queue for multi-agent dev workflows

Build queue tool for multi-agent dev workflows using Claude. Open-source developer tool for managing concurrent AI agent builds.

HN RamboZeusHammer 4/20/2026

ZeusHammer – Built an AI Agent That "Thinks Locally"

ZeusHammer: local-first AI agent combining open-source models with intent recognition, reducing API costs and latency via local processing.

HN jlewitt1 4/20/2026

Show HN: Keshro, plan and execute migrations with AI agents

Keshro: AI agent tool for planning and executing database/system migrations. Open-source developer tool leveraging agents.

HN jjgreen 4/20/2026

AI quota inflation is no token effort. It's baked in

Opinion piece on token inflation and pricing dynamics in LLM services. Analysis of LLM cost economics and quotas.

HN armcat 4/20/2026

AI Agent Traps (DeepMind)

DeepMind research on failure modes and safety challenges in AI agent design and deployment.

HN signa11 4/20/2026

C++: Growing in a world of competition, safety, and AI (herb-sutter) [pdf]

Herb Sutter PDF presentation on C++ evolution addressing competition, safety concerns, and AI integration.

HN lionkor 4/20/2026

What Claude Code Chooses

Analysis of decision-making patterns and behavior in Claude Code AI assistant.

HN sebakubisz 4/20/2026

Voicebox – The open-source voice synthesis studio

Voicebox: open-source voice cloning and speech synthesis studio with local-first processing and API.

HN nftdude2024 4/20/2026

HN: Vynly Social network for AI agents, with MCP server and demo token

Vynly social network platform for AI agents featuring MCP server integration and demo token.

HN sirnicolaz 4/20/2026

AI Agents replacing mid-management, not developers

Brief claim that AI agents will replace middle management roles rather than developer positions.

HN WavyPeng 4/20/2026

We taught AI to write correct streaming SQL

RisingWave releases official Agent Skills for AI coding agents to write streaming SQL, compatible with Claude, Copilot, Cursor, and 18+ other agents.

HN Oaklight 4/20/2026

Show HN: LLM-Rosetta - Translate LLM API Calls Across OpenAI, Anthropic, Gemini

Tool translating LLM API calls across OpenAI, Anthropic, Gemini via shared IR. Open-source adapter solving multi-provider integration without unified client.

Ax Yangyi Fang, Jiaye Lin, Xiaoliang Fu, Cong Qin, Haolin Shi 4/20/2026

Placing Puzzle Pieces Where They Matter: A Question Augmentation Framework for Reinforcement Learning

Question augmentation framework for reinforcement learning with LLMs that strategically places hints to balance easy/hard problem training.

Ax Erchi Wang, Pengrun Huang, Eli Chien, Om Thakkar, Kamalika Chaudhuri, Yu-Xiang Wang, Ruihan Wu 4/20/2026

DPrivBench: Benchmarking LLMs' Reasoning for Differential Privacy

DPrivBench investigates whether LLMs can reason about differential privacy algorithm design and verification.

Ax Jeremy Qin, Maksym Andriushchenko 4/20/2026

QuantSightBench: Evaluating LLM Quantitative Forecasting with Prediction Intervals

QuantSightBench evaluates LLM reasoning for quantitative forecasting with prediction intervals across economics and public health domains.

Ax Tristan Kirscher (ICube), Alexandra Ertl (DKFZ), Klaus Maier-Hein (DKFZ), Xavier Coubez (ICANS), Philippe Meyer (ICube), Sylvain Faisan (ICube) 4/20/2026

TwinTrack: Post-hoc Multi-Rater Calibration for Medical Image Segmentation

TwinTrack framework for post-hoc calibration of medical image segmentation models under annotator disagreement.

Ax Yaohong Yang, Sammie Katt, Samuel Kaski 4/20/2026

Multi-Objective Bayesian Optimization via Adaptive \varepsilon-Constraints Decomposition

STAGE-BO for multi-objective Bayesian optimization with adaptive constraints decomposition for expensive black-box functions.

Ax Jean-Baptiste Escudi\'e, Benjamin Barnes, Stefan Meisegeier, Klaus Kraywinkel, Fabian Prasser, Nils K\"orber 4/20/2026

Evaluating quality in synthetic data generation for large tabular health datasets

Evaluation of synthetic data generation models on large health datasets comparing machine learning families with hyperparameter tuning.

Ax Guransh Singh 4/20/2026

AEGIS: Anchor-Enforced Gradient Isolation for Knowledge-Preserving Vision-Language-Action Fine-Tuning

AEGIS framework for fine-tuning vision-language models for robotic control while preserving pre-trained knowledge.

Ax Stefano Colamonaco, David Debot, Pietro Barbiero, Giuseppe Marra 4/20/2026

Prototype-Grounded Concept Models for Verifiable Concept Alignment

Prototype-Grounded Concept Models that improve interpretability by grounding learned concepts in visual prototypes for verification.

Ax Weijiang Xiong, Robert Fonod, Nikolas Geroliminis 4/20/2026

Unveiling Stochasticity: Universal Multi-modal Probabilistic Modeling for Traffic Forecasting

Probabilistic approach for traffic forecasting addressing uncertainty and stochasticity in spatio-temporal prediction.

Ax Alina Deriyeva, Jesper Dannath, Benjamin Paassen 4/20/2026

SCRIPT: Implementing an Intelligent Tutoring System for Programming in a German University Context

Intelligent tutoring system for Python programming education using generative models to provide hints and feedback to students.

Ax Fernando Moro, Vinicius M. A. Souza 4/20/2026

Univariate Channel Fusion for Multivariate Time Series Classification

Univariate Channel Fusion method for efficient multivariate time series classification on low-cost hardware and IoT devices.

Ax Karim K. Ben Hicham, Jan G. Rittig, Martin Grohe, Alexander Mitsos 4/20/2026

Tabular foundation models for in-context prediction of molecular properties

Tabular foundation models for molecular property prediction without task-specific fine-tuning, enabling in-context learning for drug discovery applications.

Ax Minchul Kang, Changyong Shin, Jinwoo Jeong, Hyunho Lee, Younghun Go, Gyeongmin Kim, Gyeongsik Yang, Chuck Yoo 4/20/2026

Training Time Prediction for Mixed Precision-based Distributed Training

Research on predicting training time in distributed deep learning with mixed precision settings, addressing resource allocation and job scheduling.

Ax Alexandra Dragomir, Ioana Pintilie, Antonio Barbalau, Marius Dragoi, Florin Brad, Cristian Daniel Paduraru, Alexandru Tifrea, Elena Burceanu, Radu Tudor Ionescu 4/20/2026

JumpLoRA: Sparse Adapters for Continual Learning in Large Language Models

JumpLoRA framework enabling continual learning in LLMs via adaptive sparse LoRA adapters mitigating catastrophic forgetting.

Ax Yide Ran, Jianwen Xie, Minghui Wang, Wenjin Zheng, Denghui Zhang, Chuan Li, Zhaozhuo Xu 4/20/2026

Sketching the Readout of Large Language Models for Scalable Data Attribution and Valuation

RISE method for scalable data attribution and valuation in LLMs using sketching to approximate readout influence.

Ax Songtao Wang, Quang Hieu Pham, Fangcong Yin, Xinpeng Wang, Jocelyn Qiaochu Chen, Greg Durrett, Xi Ye 4/20/2026

Detecting and Suppressing Reward Hacking with Gradient Fingerprints

Method using gradient fingerprints to detect and prevent reward hacking in RL-trained LLMs without constraining intermediate reasoning.

Ax Habibeh Naderi, Behrouz Haji Soleimani, Stan Matwin 4/20/2026

Joint-Centric Dual Contrastive Alignment with Structure-Preserving and Information-Balanced Regularization

Multimodal framework (HILBERT) for learning document-level audio-text representations from long sequences in low-resource settings.

Ax Sarthak Mittal, Leo Gagnon, Guillaume Lajoie 4/20/2026

Beyond Distribution Sharpening: The Importance of Task Rewards

Research comparing whether task-reward-based RL develops new LLM capabilities or sharpens existing distribution for agent behavior.

Ax Shriram Chennakesavalu, Kirill Shmilovich, Hayley Weir, Colin Grambow, John Bradshaw, Patricia Suriana, Chen Cheng, Kangway Chuang 4/20/2026

Evaluating the Progression of Large Language Model Capabilities for Small-Molecule Drug Design

Benchmark suite evaluating LLM capabilities for small-molecule drug design across property prediction, representation, and generative tasks.

Ax Z. Cheng, N. Song 4/20/2026

When the Loop Closes: Architectural Limits of In-Context Isolation, Metacognitive Co-option, and the Two-Target Design Problem in Human-LLM Systems

Autoethnographic case study of prompt-engineering system for cognitive self-regulation, documenting behavioral changes from LLM use.

Ax Varad Vishwarupe, Ivan Flechais, Nigel Shadbolt, Marina Jirotka 4/20/2026

To LLM, or Not to LLM: How Designers and Developers Navigate LLMs as Tools or Teammates

Qualitative study of how designers and developers integrate LLMs into workflows, examining tool vs. teammate roles.

Ax Venkata Abhinandan Kancharla 4/20/2026

Applied Explainability for Large Language Models: A Comparative Study

Comparative study of explainability techniques (Integrated Gradients, Attention Rollout, SHAP) applied to fine-tuned DistilBERT for sentiment analysis.

Ax Kieran A. Murphy 4/20/2026

InfoChess: A Game of Adversarial Inference and a Laboratory for Quantifiable Information Control

Symmetric adversarial game framework for studying information acquisition and inference without material incentives or piece capture.

Ax Afia Farjana, Zaiyu Cheng, Antonio Mastropaolo 4/20/2026

Prompt-Driven Code Summarization: A Systematic Literature Review

Systematic literature review of LLM-based code summarization techniques for automatic software documentation generation using prompt engineering.

Ax Zhitong Gao, Parham Rezaei, Ali Cy, Mingqiao Ye, Nata\v{s}a Jovanovi\'c, Jesse Allardice, Afshin Dehghan, Amir Zamir, Roman Bachmann, O\u{g}uzhan Fatih Kar 4/20/2026

(1D) Ordered Tokens Enable Efficient Test-Time Search

Study on ordered tokenization enabling efficient test-time search in autoregressive generative models through token structure optimization.

Ax Jevin Jiang, Ying Chen, Blake A. Hechtman, Fenghui Zhang, Yarong Mu 4/20/2026

Ragged Paged Attention: A High-Performance and Flexible LLM Inference Kernel for TPU

Optimized LLM inference kernel for TPU deployment with efficient paged attention mechanism handling dynamic ragged workloads.

Ax Yuriy Kim, Evgeny Belyaev 4/20/2026

Lossless Compression via Chained Lightweight Neural Predictors with Information Inheritance

Lossless compression approach using chain of lightweight neural predictors for probability estimation in Markov sources.

Ax Guy Kaplan, Zorik Gekhman, Zhen Zhu, Lotem Rozner, Yuval Reif, Swabha Swayamdipta, Derek Hoiem, Roy Schwartz 4/20/2026

Why Fine-Tuning Encourages Hallucinations and How to Fix It

Study on how supervised fine-tuning increases LLM hallucinations through exposure to new facts and mitigation using continual learning techniques.

Ax Rafael T. Sereicikas, Pedro R. Pires, Gregorio F. Azevedo, Tiago A. Almeida 4/20/2026

Learning Behaviorally Grounded Item Embeddings via Personalized Temporal Contexts

Item embedding method for recommender systems that captures temporal dynamics of user preferences beyond bag-of-items approaches.