2/16/2026
Show HN: MLX-Ruby – Ruby Bindings for Apple's MLX ML Framework
Ruby bindings for Apple's MLX machine learning framework, enabling ML development in Ruby with native C++ extension wrapping MLX runtime.
2/16/2026
Show HN: Gulama – Security-first open-source AI agent (OpenClaw alternative)
Gulama: security-focused open-source AI agent with 15+ security mechanisms including AES-256 encryption, sandboxing, and credential protection.
2/16/2026
Show HN: Open API for AI agents to search 29k+ declassified docs
Open API platform for AI agents to autonomously research 29k+ declassified U.S. government documents with no paywall or API key requirement.
2/16/2026
Cognitive Debt in AI Coding
Analysis of 'cognitive debt' in AI-assisted code where developers lose mental models of AI-generated features, making maintenance harder.
2/16/2026
Show HN: AgenticMail – Email, SMS, and multi-agent coordination for AI agents
AgenticMail: self-hosted email/SMS infrastructure for AI agents with REST API, supporting send/receive/search of real email and text messages.
2/16/2026
Show HN: LLM AuthZ Audit – find auth gaps and prompt injection in LLM apps
Static analyzer tool for LLM applications detecting common security vulnerabilities: hardcoded keys, auth gaps, prompt injection, session isolation.
2/16/2026
Show HN: AI learning paths with YouTube playlists
AI learning path generator that creates structured study guides with YouTube playlists using OpenAI API and knowledge graphs.
2/16/2026
Show HN: Wisepanel – Multi-model AI panel for decision support
Wisepanel: multi-model AI tool combining ChatGPT, Claude, Gemini, Perplexity as distinct perspectives for decision support.
2/16/2026
Show HN: SafeClaw – a way to manage multiple Claude Code instances in containers
SafeClaw: container-based dashboard for managing multiple Claude Code instances with sensible defaults and skills.
2/16/2026
Laravel Dev using Claude
Laravel developer workflow example demonstrating Claude Code plugin usage for frontend design and application development.
2/16/2026
Show HN: I track GPU rental prices – same H100 ranges from $0.80 to $3.19/HR
Real-time GPU pricing aggregator comparing H100, A100, and RTX 4090 rental costs across cloud providers.
2/16/2026
Show HN: Mindweave – AI-powered personal knowledge hub with semantic search
Mindweave: semantic search tool for personal knowledge management using embeddings to find content by meaning.
2/16/2026
OpenReview MCP server with Cursor integration
MCP server providing OpenReview integration for fetching papers from ML conferences (ICML, ICLR, NeurIPS) and parsing PDFs for Claude Code.
2/16/2026
Building a multi-model AI deliberation platform: what happens when 4 LLMs interact in distinct roles
Wisepanel multi-model LLM deliberation platform where ChatGPT, Claude, Gemini, and Perplexity interact in distinct roles for complex decisions.
2/16/2026
GT-HarmBench: Benchmarking AI Safety Risks Through the Lens of Game Theory
GT-HarmBench: benchmark of 2,009 game-theoretic scenarios for evaluating multi-agent AI safety risks in frontier AI systems.
2/16/2026
A Theoretical Framework for Adaptive Utility-Weighted Benchmarking
Theoretical framework for adaptive utility-weighted benchmarking in machine learning and LLM evaluation systems.
2/16/2026
Evolving Beyond Snapshots: Harmonizing Structure and Sequence via Entity State Tuning for Temporal Knowledge Graph Forecasting
Entity State Tuning method for temporal knowledge graph forecasting that maintains entity state across timestamps to improve long-term dependency modeling.
2/16/2026
Intent-Driven Smart Manufacturing Integrating Knowledge Graphs and Large Language Models
Framework integrating instruction-tuned LLMs (Mistral-7B) with ontology-aligned knowledge graphs for intent-driven interaction in manufacturing ecosystems.
2/16/2026
Scaling Web Agent Training through Automatic Data Generation and Fine-grained Evaluation
Scalable pipeline for auto-generating training data for web agents with constraint-based evaluation framework for fine-grained progress assessment toward task completion.
2/16/2026
To Mix or To Merge: Toward Multi-Domain Reinforcement Learning for Large Language Models
Study on multi-domain reinforcement learning strategies for LLMs, comparing approaches for combining RL-verified reasoning across different expert domains.
2/16/2026
Can I Have Your Order? Monte-Carlo Tree Search for Slot Filling Ordering in Diffusion Language Models
McDiffuSE framework using Monte Carlo Tree Search to optimize slot infilling order in masked diffusion models for code and math reasoning tasks.
2/16/2026
GeoAgent: Learning to Geolocate Everywhere with Reinforced Geographic Characteristics
RL-based geolocation model using reinforcement learning and geographic characteristics for fine-grained address prediction with improved interpretability.
2/16/2026
AI Agents for Inventory Control: Human-LLM-OR Complementarity
Hybrid approach combining LLM agents with operations research algorithms for inventory control, demonstrating complementarity between flexible reasoning and formal optimization.
2/16/2026
Think Fast and Slow: Step-Level Cognitive Depth Adaptation for LLM Agents
Adaptive cognitive depth framework for LLM agents that varies reasoning depth per step, improving efficiency for multi-turn decision-making tasks with varying cognitive demands.
2/16/2026
Evaluating Robustness of Reasoning Models on Parameterized Logical Problems
Diagnostic benchmark using parameterized 2-SAT problems to evaluate LLM reasoning robustness, separating surface difficulty from structural satisfiability factors.
2/16/2026
SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks
Benchmark of 86 tasks across 11 domains evaluating how well agent skills (procedural knowledge packages) improve LLM agent performance under different conditions.
2/16/2026
WebClipper: Efficient Evolution of Web Agents with Graph-based Trajectory Pruning
Framework compressing web agent trajectories via graph-based pruning to improve search efficiency in complex information-seeking tasks.
2/16/2026
BrowseComp-$V^3$: A Visual, Vertical, and Verifiable Benchmark for Multimodal Browsing Agents
Visual and verifiable benchmark for multimodal browsing agents evaluating planning, tool-use, and deep search capabilities on web tasks.
2/16/2026
Language-Guided Invariance Probing of Vision-Language Models
Benchmark measuring vision-language model invariance to paraphrases and sensitivity to semantic changes in image-text matching.
2/16/2026
Peak + Accumulation: A Proxy-Level Scoring Formula for Multi-Turn LLM Attack Detection
Scoring formula for detecting multi-turn LLM prompt injection attacks without invoking LLM at proxy layer level.
2/16/2026
A Lightweight LLM Framework for Disaster Humanitarian Information Classification
Lightweight framework using parameter-efficient fine-tuning for disaster humanitarian information classification from social media.
2/16/2026
From Biased Chatbots to Biased Agents: Examining Role Assignment Effects on LLM Agent Robustness
Empirical study examining how demographic persona assignments introduce biases in LLM agent task performance and robustness.
2/16/2026
Retrieval-Augmented Self-Taught Reasoning Model with Adaptive Chain-of-Thought for ASR Named Entity Correction
Retrieval-augmented LLM method with adaptive chain-of-thought reasoning for correcting named entity errors in ASR outputs.
2/16/2026
Energy-Aware Reinforcement Learning for Robotic Manipulation of Articulated Components in Infrastructure Operation and Maintenance
Energy-aware reinforcement learning for robotic manipulation of articulated components in infrastructure maintenance operations.
2/16/2026
Adaptive traffic signal control optimization using a novel road partition and multi-channel state representation method
Deep reinforcement learning approach using DQN and PPO for adaptive traffic signal control optimization with multi-channel state representation.
2/16/2026
OptiML: An End-to-End Framework for Program Synthesis and CUDA Kernel Optimization
End-to-end framework using LLMs for synthesizing and optimizing high-performance CUDA kernels from natural language specifications.
2/16/2026
Perceptual Self-Reflection in Agentic Physics Simulation Code Generation
Multi-agent framework for physics simulation code generation from natural language with perceptual self-reflection validation mechanism.
2/16/2026
AgenticShop: Benchmarking Agentic Product Curation for Personalized Web Shopping
Benchmark evaluating agentic systems for personalized e-commerce product curation and web shopping automation tasks.
2/16/2026
ForeAct: Steering Your VLA with Efficient Visual Foresight Planning
Visual foresight planning system guiding vision-language-action models with imagined future observations for open-world robotic tasks.
2/16/2026
Policy4OOD: A Knowledge-Guided World Model for Policy Intervention Simulation against the Opioid Overdose Crisis
Knowledge-guided world model using policy intervention simulation for evaluating opioid epidemic mitigation strategies.
2/16/2026
Value Bonuses using Ensemble Errors for Exploration in Reinforcement Learning
Reinforcement learning exploration method using ensemble errors to compute value bonuses for directed agent exploration.
2/16/2026
Why Deep Jacobian Spectra Separate: Depth-Induced Scaling and Singular-Vector Alignment
Theoretical study of deep Jacobian spectra behavior in gradient-based training, explaining implicit bias through depth-induced singular value scaling.
2/16/2026
Rational Neural Networks have Expressivity Advantages
Theoretical analysis showing rational activation functions provide expressivity and parameter efficiency advantages over standard activations like ReLU and SiLU.
2/16/2026
What does RL improve for Visual Reasoning? A Frankenstein-Style Analysis
Analysis of what visual reasoning capabilities reinforcement learning improves versus supervised fine-tuning in vision-language models.
2/16/2026
AstRL: Analog and Mixed-Signal Circuit Synthesis with Deep Reinforcement Learning
Deep reinforcement learning method for automating analog and mixed-signal circuit design optimization across diverse, non-differentiable design spaces.
2/16/2026
Soft Contamination Means Benchmarks Test Shallow Generalization
Study on soft contamination in LLM training data showing n-gram decontamination filters miss semantic duplicates, biasing benchmark generalization estimates.
2/16/2026
CacheMind: From Miss Rates to Why -- Natural-Language, Trace-Grounded Reasoning for Cache Replacement
Conversational RAG tool using LLMs for CPU cache replacement analysis and semantic reasoning on trace data.
2/16/2026
RankLLM: Weighted Ranking of LLMs by Quantifying Question Difficulty
Benchmark framework that quantifies question difficulty to better differentiate LLM capabilities in evaluation.
2/16/2026
Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
Comprehensive framework for modular LLM agents with composable skills, covering architecture, acquisition, security considerations.
2/16/2026
Safe Reinforcement Learning via Recovery-based Shielding with Gaussian Process Dynamics Models
Safe RL framework using Gaussian process dynamics and recovery-based shielding for provably safe control.