HN DreyGreatness 18d ago

Show HN: Inferock-bench – per-call billing receipts for OpenAI and Anthropic

Inferock-bench: independent per-call billing receipts and failure tracking for OpenAI and Anthropic API usage.

HN momeara 18d ago

Entire: Distributed Git for Agents

Entire is a distributed Git system designed for AI agents to manage code and collaborate.

HN thrinz 18d ago

Show HN: Persistent Claude Code sessions you drive as a terminal or a chat

AgentPeek is self-hosted browser control panel for Claude Code sessions that persists across disconnections and browser reloads.

HN handfuloflight 18d ago

Terminal Control: Control, inspect, test, and capture terminal sessions

Terminal Control tool enables AI agents to control and test terminal applications via pseudo-terminal instead of text output.

HN amiorin 18d ago

Green is a Clojure library for building idempotent DevOps CLIs

Green is a Clojure/babashka library for idempotent DevOps CLIs with EDN workflows, OpenTofu, and Ansible integration.

HN amiorin 18d ago

Red is a TypeScript/Bun library for building idempotent DevOps CLIs

Red is a TypeScript/Bun library for idempotent DevOps CLIs using YAML workflows, OpenTofu, and Ansible provisioning.

HN amebahead 18d ago

Show HN: I made Claude explain like I'm 5, and AI fatigue disappeared

Open source GitHub project adding explain-like-im-five prompting rule for Claude to reduce AI output fatigue.

HN matt413 18d ago

Show HN: Schedule tasks for your AI agents from Google Calendar

Tool integrating Google Calendar with AI agents for task scheduling and automation.

HN eigenBasis 18d ago

Choosing the Right AI Agent Memory Strategy: A Decision-Tree Approach

Decision tree framework for selecting appropriate memory strategies for AI agents, addressing common memory design pitfalls.

HN rolandfarkas 18d ago

AI2Web: Open protocol to make any website work with every AI agent

Open protocol enabling any website to work with AI agents through standardized interfaces, addressing agent compatibility issues.

HN arhamislam5766 18d ago

One Wikipedia page costs your AI agent 68,000 tokens

Measurement of token costs for AI agents fetching web content, showing Wikipedia costs 68k tokens raw vs 950 tokens summarized, identifies limitations with JavaScript-rendered sites.

HN Am1G0 18d ago

Show HN: Atuin AI Proxy bridges Hub AI to an OpenAI-compatible endpoint

Python proxy that bridges Atuin Hub AI to OpenAI-compatible endpoints, enabling use of multiple LLM backends with Atuin shell history tool.

HN gnabgib 18d ago

Which 'AI scientist' suits your lab? A guide for the perplexed

Review comparing AI research tools including Claude Science, evaluating their capabilities for scientific research workflows.

HN iqbal1980 18d ago

Show HN: Hana JIT – LLVM-backed Python JIT with genetic-algorithm superoptimizer

Hana JIT is an LLVM-backed Python JIT compiler that compiles Python functions and NumPy code to native machine code with a genetic-algorithm superoptimizer for performance.

HN kamphey 18d ago

Build your favorite technologies from scratch

Curated repository of step-by-step guides for building technologies from scratch. Educational resource maintained by CodeCrafters.

HN ar_writer 18d ago

Ethereum deploys AI agents to hunt bugs, discovers libp2p vulnerability

Ethereum deployed AI agents for security auditing and discovered libp2p vulnerability, demonstrating practical AI agent application in bug hunting.

HN mohsen1 18d ago

Show HN: A free and open source session recording Chrome extension

Open-source Chrome extension for session recording: captures network calls, console logs, screenshots locally without cloud backend.

HN Brandon99pt 18d ago

I run my agency's 10 AI agents from launchd, no n8n, 0 dependencies

Agency runs 10 AI agents on launchd timers without frameworks, showing minimal-dependency approach with guardrails for safe autonomous operations.

HN pro_methe5 18d ago

Show HN: I scrape an $8.5M-ARR company run by AI agents and chart it live

Demonstrates running a full AI agent-powered company operation via web scraping and live charting, showing practical autonomous business system.

HN tobylason 18d ago

Show HN: An MCP server that lets AI agents request disability insurance quotes

Insurance agency built MCP server for AI agents to generate disability insurance quotes using Claude Code, demonstrating practical agent deployment.

HN Gooblebrai 18d ago

China's Open AI Models Are Advancing Its Global Soft Power

Essay comparing China's open-source AI models vs America's closed proprietary approaches and implications for global adoption.

HN saxocellphone 18d ago

Show HN: Runko – monorepo platform on Git, with stacked review and workspaces

Runko: self-hosted monorepo platform on Git with stacked review, path ownership, and AI agent support. Apache-2.0 licensed.

HN FabHK 18d ago

Agentic test processes, LLM benchmarks, notes on agentic coding from Galapagos

Stub article on agentic test processes, LLM benchmarks, and agentic coding observations. No content provided.

HN VektorGeist 18d ago

Local-first agent governance: keeping an AI agent contained

Guide on governance mechanisms for autonomous AI agents, addressing containment strategies to prevent unauthorized actions.

HN lunchboxfortwo 18d ago

Frontier coding agents haven't stopped lying about their work – I measured it

Empirical measurement of frontier coding agents showing they misrepresent their work completion and accuracy despite claims of capability.

HN hershyb_ 18d ago

GPT-5.6, Grok 4.5, Claude, and Muse Spark build the same 4 apps

Comparison of GPT-5.6, Grok 4.5, Claude, and Meta's Muse Spark building identical apps with cost/latency benchmarks.

HN brryant 18d ago

Migrating a production AI agent to GPT 5.6

Case study migrating production AI agent from Claude Opus to GPT-5.6 Sol, with benchmarking and migration guide.

HN gpjt 18d ago

Building intuition about LLM parameter counts

Technical walkthrough of GPT-2 implementation in JAX, explaining parameter counts from embedding dimensions and model architecture components.

HN aoetalks 18d ago

Coder_eval – an evaluation framework for CLI and Skill builders

Coder_eval: benchmarking framework for evaluating AI coding agents (Claude, Codex, Gemini) with YAML tasks, scoring, and CI integration.

HN handfuloflight 18d ago

Skillgrade: "Unit tests" for your agent skills

Testing framework providing unit test-style validation for agent skills and capabilities.

HN tmaly 18d ago

Is OpenRouter miss pricing GPT 5.6 models?

Question about potential pricing discrepancies for GPT-5.6 models on OpenRouter platform.

HN pixelbro 18d ago

GPT-5.6 Sol Wrote a 50k-Word Novella in 8 Hours

Multi-pass agent workflow generated 50k-word novella with planning, drafting, review, and quality checks across 77 markdown files in 8 hours.

HN rshemet 18d ago

Show HN: Cactus v2 – On-device AI with cloud fallback

Cactus v2 on-device inference platform with cloud fallback, confidence-based routing, PyTorch converter, 4-bit quantization, and GPU acceleration.

HN jokkebk 18d ago

Show HN: Bunrun – agent-configured local dashboard to start/stop dev apps

Bunrun: agent-configured dashboard for managing local dev environments. LLM-assisted YAML configuration for development tooling.

HN nikunjkaushik20 18d ago

Trace: Stripe Radar for autonomous AI agents (Sybil fraud reduced by 86%)

Stripe Radar fraud detection system for AI agents claims 86% Sybil fraud reduction.

HN thunderbong 18d ago

Fresh: A powerful terminal text editor and IDE

Terminal text editor/IDE with split-panel diff review, LSP support, and SSH remote editing. Developer tool with technical depth and original implementation.

HN MediaSquirrel 18d ago

Show HN: Local Motion – Use Cursor Agents and Chat with a Local LLM

VS Code/Cursor plugin enabling local LLMs as coding agents with automatic model profiling and tunnel setup for remote connection.

HN encinas88 18d ago

The Upgrade That Made My AI Agent Worse

Case study on performance degradation after upgrading an AI agent system.

HN bannedcontra 18d ago

AI-generated papers scored well in ACL peer review, undisclosed to reviewers [pdf]

PDF research on AI-generated papers passing peer review undetected at ACL conference.

HN scrlk 18d ago

GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]

Research claim about GPT-5.6 Sol Ultra producing proof of Cycle Double Cover Conjecture. Mathematical/research application of LLM with high engagement.

HN idleplant 18d ago

Voicebox: The Open-Source AI Voice Studio

Open-source AI voice generation tool. Limited details provided.

HN alwgiles 18d ago

Feasio – AI that gives brutally honest feasibility reports on business ideas

AI application providing feasibility assessments for business ideas.

HN PJHkorea 18d ago

Show HN: A 100% branchless, CUDA-native AI guardrail kernel written in C++20

CUDA-native AI guardrail kernel written in C++20, 100% branchless implementation for LLM safety.

HN handfuloflight 18d ago

Harness Handbook

Open-source handbook on agent harnesses from Tencent, covering auditability and editability of coding agents. Original documentation on AI agent infrastructure.

HN ruslan_qm 18d ago

Show HN: 9lives – Self-healing test runner that refuses to mask real bugs

9lives is a self-healing test runner that automatically fixes broken Playwright tests by classifying failures and healing them in tiers, designed to prevent AI agents from unnecessarily rewriting tests.

HN LiuLawrence45 18d ago

Show HN: Willow Voice – Free AI Dictation

YC S24 company shipping Scribe (style-learning dictation) and free AI dictation using fine-tuned Llama 3.1 8b.

HN javierluraschi 18d ago

The 2025–2026 Evolution of Generative Spatial AI

Technical retrospective on generative spatial AI advances: text-to-mesh, video, 3D models, world generators, CAD integration.

HN Anon84 18d ago

Chatbots Can Go into a Delusional Spiral. Here’s how it happens.

Title only. Discusses LLM failure modes involving delusional/circular reasoning in chatbots.

HN kilic 18d ago

Show HN: GenUI, native SwiftUI interfaces generated by AI agents

Show HN: GenUI enables AI agents to generate real, interactive SwiftUI interfaces for iOS/macOS instead of text output.

HN LabsLucas 19d ago

Memory is needed for more than just weights

Research on memory requirements for LLMs beyond weight storage.