HN m0rvayne 25d ago

MCP server that lets Claude click menus on your Mac and fix its own mistakes

MCP server enabling Claude to interact with Mac UI, fixing mistakes autonomously.

HN heinrich5991 25d ago

The unreasonable effectiveness of LLMs for auditing Rust code

Study on using LLMs to audit Rust code for security vulnerabilities.

HN nephihaha 25d ago

Sword Swallowing and Its Side Effects

Medical study on sword swallowing complications and associated injuries.

HN train212 25d ago

POSBox – Free Open Source Point of Sale System

Open source offline point-of-sale system for small businesses.

HN TheWiggles 25d ago

Jazzy – Productive, Developer-Friendly Web Framework for Nim

Nim web framework with batteries-included features for rapid development.

HN theanonymousone 25d ago

DeepSWE Benchmark updated with GLM 5.2 and updated results for other models

DeepSWE benchmark for measuring long-horizon coding agents on engineering tasks, updated with GLM 5.2 results.

HN cromka 25d ago

A controlled hypothermia cure for Lyme disease [video]

Medical article on Lyme disease treatment using controlled hypothermia.

HN JTR85 25d ago

A neuro-adaptive OS concept for energy efficiency and execution-path attestation

Conceptual paper on neuro-adaptive OS using neural models for energy efficiency and execution path attestation.

HN swolpers 25d ago

The AI Definition of Done: Human in the Loop Is Not a Quality Standard

Framework for establishing quality standards for AI-generated outputs in team workflows.

HN 44za12 25d ago

Horcrux – Distributed, Zero-Trust Secret Manager

Distributed secret manager using Shamir's Secret Sharing and Reed-Solomon codes across cloud providers.

HN birdculture 25d ago

I am dreading our LLM-written incident report future

Essay expressing concerns about future LLM-written incident reports replacing human analysis.

HN grisun0 25d ago

FreeDom free and open-source minimal web browser – Zero Trust and Zero Knowledge

Minimal open-source web browser written in C emphasizing privacy, zero-knowledge, and zero-trust architecture.

HN jingkai_he 25d ago

Maco – Let your agent rg a filesystem of MCP tools and run them as code

MCP-as-code tool that collapses multiple MCP servers into lean Pydantic/Python interface for AI agents.

HN tomaskafka 25d ago

Open Ralph Wiggum – Autonomous Agentic Loop

Open-source autonomous agentic loop supporting Claude, OpenAI, Copilot with sandboxed execution via sandboxed.sh.

HN lum1104 25d ago

Agentic Systems Course: Learn AI Agents with an AI Coding Agent

22-chapter course on designing, building, and operating production AI agents with autonomous goal-pursuit capabilities.

HN Byron230686 25d ago

Beast: Inference Economy Inversion in Agentic Coding Systems

BEAST system for safer agentic coding by intercepting agent actions, verifying results, and bounding model outputs.

HN bashierkh 25d ago

Governing AI-agent actions via a network intent layer (NILScript)

NILScript network intent layer for governing AI agent actions. Control mechanism for agent execution.

HN Garbage 25d ago

Agent-trace: A standard format for tracing AI-generated code

Standard format for tracing AI-generated code. Infrastructure for monitoring AI agent behavior.

HN david3289 25d ago

I made an Time-Based AI coding agent

Time-based AI coding agent with web IDE, terminal access, and GitHub integration. Open-source developer tool.

HN sarangk90 25d ago

Building Reliable Agentic AI Systems

PRINCE platform case study: Bayer/Thoughtworks built production agentic RAG system with Text-to-SQL for pharmaceutical research, integrating decades of safety reports into intelligent assistant.

HN sjashwin 25d ago

Show HN: Moduna –- Mixpanel for AI Agents

Moduna provides observability and decision intelligence for production AI agents, turning conversations into structured product insights and failure analysis.

HN asxndu 25d ago

Ask HN: How to get ideas for space startups?

Discussion thread asking for space startup ideas. Off-topic for AI/ML interests.

HN titusblair 25d ago

Show HN: Argybargy – A peer-to-peer bridge connecting any AI agents and sessions

Argybargy is peer-to-peer bridge enabling multiple AI agents/sessions to communicate and coordinate across machines, apps, and model vendors via HTTP without SDK dependencies.

HN Oxlamarr 25d ago

A TypeScript prototype admission layer for agents

Attestor is an admission layer for AI agents that sits between agent operations and system execution, preventing unsafe/unauthorized service calls through control infrastructure.

HN eaftan 25d ago

SafeRE: Building a production-quality regex library with agents

Building production-quality regex library using AI agents. SafeRE series demonstrates using frontier models for complex long-running software development tasks.

HN hi2brain 25d ago

Second Brain – A free, invisible AI interview copilot (Groq and Llama 3)

Second Brain: free desktop copilot using Groq/Llama-3 for real-time interview assistance via voice capture and context-aware suggestions from resume/job description.

HN vantareed 25d ago

Palmier-Pro: macOS video editor built for AI

Palmier Pro: open-source Swift-based macOS video editor enabling AI agents to generate/edit videos via integrations with Claude/Codex and MCP protocol.

HN vinhnx 25d ago

Train LLM from Scratch

Complete guide to training transformers from scratch in PyTorch without frameworks. Covers post-training, instruction tuning, reasoning on single/multi-GPU setups with public datasets.

HN adltereturn 25d ago

Show HN: Agent-historian – let your AI agent search its own past sessions

Agent-historian enables AI coding agents to search past conversation sessions from CLI, allowing agents to recover previous research and avoid redundant work across stateless sessions.

HN random__duck 25d ago

Moving Machine Learning into the Analog Domain

Abstract discussion on computing paradigms, analog systems, and hardware constraints. Incomplete content.

HN abipal15 25d ago

Plotting AI model release cadence: two labs are accelerating, three aren't

Data-driven analysis of AI lab model release cadence. Tests hypothesis that labs with self-improvement capability ship models faster using publicly available release timeline data.

HN ethanhq 25d ago

Show HN: Cc-fleet – run other LLMs as Claude Code workers, your sub drives

cc-fleet bridges third-party LLMs to Claude Code's multi-agent orchestration, enabling any API-compatible model to run as subagents. Open source integration tool.

HN JackLau 25d ago

Ratchet – BIOS flashing toolkit with a built-in MCP server for AI agents

Hardware debug toolkit with MCP server integration enabling AI agents to perform BIOS flashing and silicon analysis across multiple protocols.

HN vinhnx 25d ago

How the AI Village Works

AI Village dataset released on HuggingFace: multi-agent trajectories of groups pursuing long-horizon goals like research and competition with computer access and internet connectivity.

HN vnbrs 25d ago

When I reject AI code even if it works

Brief post title only. Discusses rejection of AI-generated code despite functionality.

HN aniketwattawmar 25d ago

Show HN: GenAIDojo

Interview prep platform for AI engineers. Practice problems on multi-agent systems, RAG, vector databases, and production AI architectures.

HN baddash 25d ago

Subquadratic claims it broke through a bottleneck that's holding back LLMs

Subquadratic claims breakthrough in mathematical bottleneck limiting LLM scaling. Startup shares independent evaluation results but skepticism remains.

HN msradam 25d ago

Show HN: Rocannon – Any Ansible module to MCP Tool, record sessions as playbooks

Tool converting Ansible modules to MCP tools for LLMs with session recording as playbooks. Leverages ansible-doc schema for structured tool calling.

HN breve 25d ago

Adobe adds its AI assistant to Premiere, Illustrator, and InDesign

Adobe expands Firefly AI assistant across Premiere, Illustrator, InDesign with new capabilities for video, branding, and asset management.

HN pro_methe5 25d ago

The History of TypeScript

Visualization and history of TypeScript language evolution from 2014-2021, covering 623 contributors and 279K lines of code.

HN gerardordz96 25d ago

Show HN: Rlsgate – Block the Supabase RLS leak before you deploy (CLI)

Rlsgate: CLI tool blocking Supabase Row Level Security vulnerabilities before deployment, addressing patterns in AI-generated applications.

HN alexreysa 25d ago

Show HN: Agentic coding workflows built on Git worktrees and task evidence

glueRun-go: Bash/Python orchestration engine for autonomous multi-agent AI coding with three-tier scheduling, git-worktree isolation, and state management.

HN mirkofr 25d ago

Show HN: FERNme – agent memory that updates with ~zero LLM calls

Graph-based persistent memory system for AI agents using fuzzy edges and Hebbian learning. Reduces LLM token usage for personalized agent responses.

HN gaggle_dk 25d ago

Ask HN: After you ship a feature, what happens to what you learned?

Discussion about capturing and surfacing learnings from shipped features to prevent repeating mistakes in development.

HN ahvural 25d ago

Show HN: Vitrus – the company brain that tells you what it doesn't know

Vitrus: AI knowledge system returning answers with sources and explicit gaps, readable by both humans and agents in Markdown.

HN jadamsl 25d ago

Hackingpal

HackingPal: Open-source AI-assisted security workbench for authorized penetration testing with engagement tracking and audit trails.

HN ahriad 25d ago

Ask HN: What are your parameter count estimates for Opus 4.8 and GPT-5.5?

Discussion thread asking community estimates for parameter counts of unreleased frontier AI models.

HN arishaller 25d ago

Show HN: Callimachus – Local search across your AI coding-agent history

Local search tool for AI coding-agent conversation history across 11 platforms. Includes MCP server, CLI, and VS Code extension. Privacy-focused.

HN bingewave 25d ago

One Prompt Agentic AI Marketing for Game Developers

Incomplete stub about agentic AI for marketing to game developers.

HN kareldecherf 25d ago

Show HN: Maccha – Cross Agent Brain for Antigravity, Claude Code, OpenCode etc.

MACCHA: Multi-agent context harness with persistent memory using vector embeddings and semantic conflict detection for AI coding agents like Claude Code and OpenCode.