5/13/2026
Claude Code weekly limits increasing 50% till July 13
Anthropic increases Claude Code weekly usage limits by 50% through July 13.
5/13/2026
Show HN: Open-source clone of the swarm/ctd/happenstance for warm intros
Open-source tool to find warm introductions from contact networks by matching LinkedIn profiles, alternative to paid services.
5/13/2026
Red Hat's skill packs give AI agents 20 years of institutional memory
Red Hat's skill packs enable AI agents with institutional memory access. Agent tooling feature unclear.
5/13/2026
Is Google or Exa better for RLing agents?
Research comparing Exa vs Google as search backends for reinforcement learning trained agents, showing RL performance improvements.
5/13/2026
Build a Basic AI Agent from Scratch
Tutorial series building AI agents from scratch to explain how agents work, covering Claude, Codex, and related tools.
5/13/2026
Apple Working on Plan to Allow AI Agent Apps on the App Store
News that Apple is developing policies to allow AI agent applications on App Store.
5/13/2026
Predicting Rare LLM Failures with 30× Fewer Rollouts
Research on predicting rare LLM failures using 30× fewer rollouts for testing efficiency.
5/13/2026
I found 500 vulnerabilities in public apps using LLMs, now what?
Security researcher found 500 vulnerabilities in apps using LLMs for detection. Title only, minimal details.
5/13/2026
Show HN: Sunex Optics – MCP server for choosing best lens/CMOS image cameras
MCP server for Claude/ChatGPT providing access to lens/CMOS camera dataset for optics selection in AI applications.
5/13/2026
How LLMs Work
Technical explanation of how language models work mechanically, covering transformer internals and token prediction.
5/13/2026
Krea launches K2, an image model built for aesthetics
Announcement of K2 image generation model optimized for aesthetic output.
5/13/2026
Tau-knowledge: benchmarking agents on real-world knowledge
Benchmark for evaluating customer-facing AI agents on real-world knowledge base navigation and application tasks.
5/13/2026
VibeLoom – contract-driven AI coding
VibeLoom tool for AI-assisted code generation using contract-driven development.
5/13/2026
Learning, Fast and Slow: LLMs That Adapt Continually
Research on LLMs that combine fast and slow learning mechanisms for continual adaptation.
5/13/2026
Show HN: Profine – Automated profiling and code rewrites for ML training loops
Profine tool for automated profiling and code rewriting optimization in ML training loops.
5/13/2026
Journey building an AI coding agent for writing and maintaining integrations
Engineering case study of AI coding agent that opens production pull requests for payment integrations, addressing undocumented quirks.
5/13/2026
Show HN: AgentDeck – a game console for AI agent research
AgentDeck: game console environment for AI agent research and testing.
5/13/2026
Launch HN: Ardent (YC P26) – Postgres sandboxes in seconds with zero migration
Ardent (YC) builds database sandboxes for coding agents to test safely without production risk.
5/13/2026
Agent pull requests are everywhere
Guide to reviewing agent-generated pull requests, technical debt patterns, and code quality issues in AI-generated code.
5/13/2026
Archivists Turn to LLMs to Decipher Handwriting at Scale
Archivists using LLMs to decipher handwritten historical documents at scale for digital preservation.
5/13/2026
Thoughts on Claude Code 2.1.139 Agent View and Background Sessions
User review of Claude Code 2.1.139's new Agent View and background sessions, noting useful features with remaining rough edges.
5/13/2026
Audrey: Local-first memory guard for AI agents (source)
Audrey - local-first memory management system for AI agents with source code available.
5/13/2026
Most teams optimize the prompt. Agentic systems have more moving parts
Discussion of complexity in agentic systems beyond prompt optimization, noting multiple moving parts.
5/13/2026
Show HN: EleutherAI / Lm-Evaluation-Harness
EleutherAI's LM Evaluation Harness - benchmarking framework for evaluating language model performance across tasks.
5/13/2026
Show HN: BossHogg: A PostHog CLI for Agents
BossHogg is a PostHog CLI for AI agents enabling HogQL queries, feature flags, and insights from terminal or Claude Code.
5/13/2026
Show HN: Ledger – Claude Code Token Spend Analyzer
Ledger is a local Rust tool analyzing Claude Code token spend via JSONL logs with CLI and dashboard.
5/13/2026
AI is making it easy but also hard
Opinion piece on competitive challenges of building products with AI coding assistants becoming accessible to all.
5/13/2026
Show HN: Ratify Protocol – prove who authorized an AI agent, offline, in <1ms
Ratify Protocol - cryptographic method to prove AI agent authorization offline with sub-millisecond verification.
5/13/2026
Some Business Ideas
Business ideas for AI infrastructure including service connectors, CDN, and database integrations for AI assistants.
5/13/2026
Building with Claude Managed Agents – Sharp Edges
Technical overview of Anthropic's Claude Managed Agents released April 2026: hosted agent infrastructure with sandboxed environments.
5/13/2026
Burn – K8s cost waste by namespace and pod. Just kubectl, no deploy
Kubernetes cost analyzer using Claude API to identify waste by namespace/pod without agents or dashboards.
5/13/2026
Microsoft’s new multi-model agentic security system tops industry benchmark
Microsoft MDASH - multi-model agentic security system that discovered 16 vulnerabilities including 4 critical RCE flaws in Windows.
5/13/2026
Show HN: Endpoint Context Protocol – Browsers get HTML, AI agents get Markdown
Endpoint Context Protocol converts HTML to Markdown for AI agents. Developer tool for agent web interaction.
5/13/2026
Build Long-running AI agents that pause, resume, and never lose context with ADK
ADK framework for building persistent AI agents with pause/resume and context preservation. Developer tool for agent development.
5/13/2026
AgentGate – Authorization layer for AI agents
AgentGate authorization framework for AI agents. Developer tool for access control in agent systems.
5/13/2026
Scherlok – zero-config data quality monitoring, works with dbt
Scherlok data quality monitoring tool using ML to detect production data anomalies, integrates with dbt.
5/13/2026
Hydrogen jukeboxes: on the crammed poetics of "creative writing" LLMs
Critical analysis of OpenAI's creative writing LLM, examining stylistic patterns and limitations in generated text.
5/13/2026
A proposed standard for governing autonomous AI deployment
Proposed governance standard for autonomous AI systems deployment.
5/13/2026
Show HN: My tool generates 3D objects composed of separate, functional parts
Tool for generating 3D objects with separate, editable parts using LLMs (Gemini, Claude, ChatGPT) to create Blender construction scripts.
5/13/2026
Cracking Jane Street LLMs
Research code and findings from cracking Jane Street LLM models with mechanistic analysis and ablation studies.
5/13/2026
What breaks when AI-generated code reaches production systems
Analysis of production system failures when AI-generated code reaches deployment, from applied AI experts.
5/13/2026
Kyunghyuncho Teaching Fundamentals of Machine Learning
NYU course on machine learning fundamentals taught to CS seniors, updated post-ChatGPT with new teaching approaches.
5/13/2026
Show HN: Promptcellar – capture every Claude Code prompt as JSONL in your repo
Claude Code plugin that logs every prompt to JSONL in repository, creating audit trail and history of agent-generated commits.
5/13/2026
Open-source AI modding tool for Rimworld (and soon Factorio)
Modmixer is open-source AI tool for building and fixing Rimworld mods by decompiling game code and auto-fixing errors detected in logs.
5/13/2026
Show HN: Rotunda - A Browser Built for Agents
Rotunda is a Firefox fork designed for AI agents with improved efficiency for computer use models and web navigation primitives.
5/13/2026
Fine-Tuning TranslateGemma-4B for Better Welsh Translations on an H200 GPU
Engineering guide to fine-tuning TranslateGemma-4B for English-Welsh translation on H200 GPU with LoRA and GGUF deployment.
5/13/2026
Open source voice-to-text for Linux. And macOS
OSTT is terminal-native speech-to-text tool for Linux/macOS supporting OpenAI, Deepgram, Groq, and other providers with hotkey integration.
5/13/2026
Show HN: Headless Cloud Security – Headless SaaS has come to security
Security company describes agentic and headless workflows for security operations, discussing adoption of AI agents in enterprise security.
5/13/2026
I tried to put an on-device LLM in an iOS Share Extension. It didn't fit
Technical exploration of constraints when deploying on-device LLMs in iOS Share Extensions.
5/13/2026
Helm AI Kernel – an open-source execution firewall for AI agents
Open source execution firewall for AI agents providing sandboxing and safety controls for agentic AI systems.