HN allenleee 22d ago

Show HN: AirPosture – Open-Source Posture Coach Using AirPods

Open-source posture tracking app using AirPods sensors for health monitoring.

HN agulaya24 22d ago

Linux Foundation Is Pursuing Trusted Identity Infrastructure for AI Agents

Linux Foundation developing trusted identity infrastructure for AI agents. Announcement only.

HN nrig 22d ago

When is an AI agent's approval prompt a security boundary?

Analysis of approval-prompt security boundaries in open-source AI agents. Author's findings reclassified out of scope after project policy change.

HN Darkstryder 22d ago

Halvar's Guide to Entrepreneurship

Guide on entrepreneurship based on author's two company exits.

HN teepo 22d ago

Nvidia Announces BioNeMo Agent Toolkit

Nvidia BioNeMo Agent Toolkit announcement for building AI agents in biomedical domain. Limited technical detail.

HN handfuloflight 22d ago

Aside: The AI Browser

Browser tool with AI capabilities. Minimal information provided.

HN P_Economics 22d ago

The Agent-Buyer Thesis – What Happens When AI Agents Are the Buyers

Speculative thesis on economic implications of AI agents becoming market buyers. Conceptual discussion with limited detail.

HN skzv 22d ago

The Geometry of Noise: Why Diffusion Models Don't Need Noise Conditioning

Machine learning research on diffusion model geometry showing noise conditioning may be unnecessary. Technical paper announcement.

HN thelonelyborg 22d ago

We've seen AI hype cycles before. They didn't end well

Historical perspective on AI hype cycles from 1950s-present, discussing winter periods and booms. Analysis without current technical focus.

HN ASRDETYnFXDGh 22d ago

Doesitstillwork.ai – community-verified status for AI tool features

Doesitstillwork.ai: Community-verified status tracking for AI tool features. Limited technical details.

HN ZappoMan 22d ago

NanoClaw-webchat – Localhost chat for multi-agent NanoClaw forks

Local browser chat interface for multi-agent NanoClaw systems with @mentions, threading, and optional MCP server integration.

HN manveerc 22d ago

Beyond Enterprise-Managed Authorization for MCP

Proposal for enterprise authorization controls in Model Context Protocol (MCP) systems.

HN ishwantsingh 22d ago

Flight Recorder for AI Agents

Flight Recorder tool for monitoring and recording AI agent execution traces.

HN singam96 22d ago

Show HN: ReflexConv2d – 57% less blur in image reconstruction

ReflexConv2d convolutional layer using weight-based masks for spatial modulation, reducing blur in image reconstruction by 57%.

HN vorticotech 22d ago

Serving Large Language Models with a Minimalist Python CLI

Minimalist Python CLI tool for serving large language models.

HN doener 22d ago

LightOn: Production RAG without the 9-month build

LightOn platform for production RAG systems without lengthy development cycles.

HN ojowwalker77 22d ago

BonsAI – a spatial board for brain-dumping and prompting agents

macOS app for capturing ideas and creating AI-ready prompts with semantic linting to detect vague instructions for agents.

HN theanonymousone 22d ago

Wavacity: Audacity audio editor ported to WASM, under GNU GPL v2

Web port of Audacity audio editor to browser using WASM and GPL v2 license.

HN reidotdev 22d ago

Show HN: Browse design inspiration from terminal while Claude is thinking

Terminal tool displaying design inspiration from multiple sources while Claude processes requests.

HN dp12 22d ago

Show HN: Proctor – signed isolation bundles for AI coding-agent benchmarks

Proctor: Tool for creating signed isolation bundles to benchmark AI coding agents.

HN rodrigo-arenas 22d ago

Show HN: Sklearn-genetic-opt – evolutionary optimization for scikit-learn

Hyperparameter optimization library using evolutionary algorithms for scikit-learn with adaptive mutation and early stopping.

HN ngrislain 22d ago

Vibe Under Constraint - Claude Code writing Lean 4

Analysis of AI agent limitations in code generation using Claude, focusing on resource constraints and practical testing.

HN b-man 22d ago

Does AI Adoption Improve Productivity? Effects over the First Three Years

Empirical study examining whether AI adoption reduces work time and improves productivity over three years.

HN adchurch 22d ago

Show HN: Optimal model routing directly in Claude, Codex and Cursor

Drop-in proxy routing requests across Anthropic, OpenAI, and Gemini models using embedder-based selection ranked #1 on RouterArena.

HN plaidthunder 22d ago

Why smarter models won't lead to AI co-workers

Research paper on role confusion and prompt injection vulnerabilities in LLM-based agents.

HN AdilZtn 22d ago

Elenchus: The open-source Claude Tag

Open-source multiplayer workspace for sharing Claude, Gemini, ChatGPT conversations with per-user API keys and billing.

HN saikatsg 22d ago

Sakana AI Releases 'Fugu Ultra' to Match Frontier Performance

Sakana AI releases Fugu Ultra model matching frontier model performance.

HN theanonymousone 22d ago

TeXlyre: A local-first LaTeX and Typst web editor

Local-first LaTeX and Typst web editor with real-time collaboration using React, TypeScript, and Yjs CRDTs.

HN iknownthing 22d ago

Ask HN: Are people generally interested using LLMs for learning purposes?

HN discussion asking why interactive LLM-based tutoring apps haven't become popular despite initial expectations.

HN ilreb 22d ago

Genesis Workbench: Generative AI for Life Sciences

Genesis Workbench uses foundation models for drug discovery, molecular prediction, and biomedical literature analysis.

HN mlinksva 22d ago

Advocating for fixes to California AI Transparency Act to protect open source

Open source coalition calls for amendments to California AI Transparency Act to protect open source licensing.

HN matt_d 22d ago

Inference Compute Shapes Frontier LLM Evaluation

Research examining how inference-time compute allocation affects frontier LLM evaluation and performance metrics.

HN DjNero 22d ago

Show HN: I created agent skill based on Peter Lynch's books

AI agent for stock analysis trained on Peter Lynch investment books with cited sources.

HN sizzle 22d ago

AI Hiring Tools Yield Racial Bias and Systemic Rejection; 26% Black & 15% Asian

Large-scale study finds AI hiring algorithms exhibit racial bias: 26% rejection rate for Black candidates, 15% for Asian candidates versus baseline.

HN lizsteelman 22d ago

AI Didn't Make Us Faster It Demoted Us

Case study: Two designers used AI tools to build brand identity quickly for media company. Reflects on quality and decision-making concerns.

HN Jibran12345 22d ago

The first AI marketing benchmark

ViralBench benchmark tests AI agents' ability to capture attention, manipulate, and self-correct. Models run in agentic loops twice daily with cloneable repo.

HN rapiddev 22d ago

Confidence estimation is a better metric than agreement for LLM judges

Research on using confidence estimation rather than agreement metrics for evaluating LLM judges in benchmarking tasks.

HN rapiddev 22d ago

LLMs Are Digitizing Judgment

Conceptual discussion on LLMs' role in automating human judgment tasks. Limited technical depth.

HN jkanalakis 22d ago

Show HN: Appstr for Indies managing lots of apps

Appstr platform helps indie app developers manage multiple app store submissions and customer feedback.

HN gmays 22d ago

GLM-5.2 Is the New Best Open Model

GLM-5.2 open-source LLM model shows benchmark improvements over GLM-5.1, evaluated as strongest open model currently available.

HN ajdrevrel 22d ago

AI Models Soccer Tournament [video]

Video showcasing AI models competing in simulated soccer games.

HN mikepollard_dev 22d ago

Show HN: RLM-based local debugger for AI agent traces

HALO: Open-source debugging tool for AI agent traces using RLM to identify failure modes and suggest fixes.

HN douglas_waugh 22d ago

TDD is how I trust the code AI agents write [video]

Video discussing test-driven development as validation approach for code written by AI agents.

HN ethanpil 22d ago

Book-to-skill: Turn a technical book PDF into a Claude Code skill

Tool to convert technical book PDFs into Claude Code skills for AI-assisted development.

HN kishan8910 22d ago

Mission control for your Claude Code sessions

Mission control dashboard for Claude Code sessions. Minimal details on features or functionality provided.

HN alehlopeh 22d ago

Show HN: Hallu – a web framework where an LLM hallucinates your app

Web framework where LLMs generate application code and interfaces dynamically.

HN gwhyyy 22d ago

Think this open-source Flutter-native AI agent worth building?

Discussion about building an open-source Flutter-native AI agent.

HN ilreb 22d ago

PhoneBuddy: Training Open Models for Agentic Phone Use

PhoneBuddy: Training framework for open-source models to perform agentic phone interactions. Enables phone automation via AI agents.

HN bonigv 22d ago

Show HN: Grove – fast source insights for coding agents – via Tree-sitter

Grove: Fast source code analysis tool for coding agents using Tree-sitter parser. Enables agents to understand codebases quickly.

HN thm 22d ago

Anthropic rolls out Claude Tag, your new agentic AI coworker in Slack

Anthropic launches Claude Tag, an AI agent for Slack that acts as an agentic coworker.