5/22/2026
Don't just paste the AI at me
Opinion piece criticizing developers who paste unedited LLM outputs directly to users without adding value or context.
5/22/2026
Did Google's AI agents build an operating system for $916?
Google's AI agents built operating system using Gemini 3.5 Flash and Antigravity 2.0 agent framework with minimal cost and single prompt.
5/22/2026
Show HN: BonzAI – self-sovereign, local LLM inference in the browser
BonzAI enables local LLM inference in the browser with self-sovereign control. Open-source LLM application tool.
5/22/2026
MCP-safeguard: Automated security scanner for MCP servers (52 detection rules)
MCP-safeguard: Security scanner for Model Context Protocol servers with 52 detection rules for automated vulnerability scanning.
5/22/2026
High-Volume VRP Optimization at Amazon Scale on a Raspberry Pi 400
Vehicle routing optimization handling 1M stops on Raspberry Pi 400 in 10 hours, proving efficient algorithms beat cloud infrastructure.
5/22/2026
ReconForge – Open-Source Security Reconnaissance Toolkit
AI-assisted reconnaissance toolkit for security researchers combining automation with LLM triage prompts for prioritizing findings.
5/22/2026
Twelve Ways to Be Wrong About AI-Assisted Coding
Analysis of twelve flawed approaches to measuring AI-assisted coding tool effectiveness and productivity metrics.
5/22/2026
Cloudflare's Matthew Prince says AI has made a category of workers obsolete
Cloudflare CEO argues AI has made certain worker categories obsolete as company grows and expands customer base.
5/22/2026
Show HN: Microcodegen.py – PRD → FastAPI app, one file, no LLM calls
Microcodegen.py generates FastAPI applications from PRDs without LLM calls. Code generation developer tool.
5/22/2026
Measuring LLMs' ability to develop exploits
Research measuring LLMs' capability to develop security exploits. Evaluates LLM abilities on malware/exploit generation.
5/22/2026
Aimeflux – Local-first macOS dictation with searchable transcript history
Local-first macOS dictation app using Whisper transcription with searchable history and optional AI cleanup on Apple Silicon.
5/22/2026
Use Grok in OpenCode
Integration of xAI's Grok model as coding agent within OpenCode editor for X Premium subscribers.
5/22/2026
Show HN: Mneme – Open-protocol AI memory that lives on your device
Open-protocol AI memory system storing encrypted data locally on device with cross-platform compatibility across Claude, ChatGPT, Gemini. Apache 2.0 on npm.
5/22/2026
Models.dev: open-source database of AI model specs, pricing, and capabilities
Open-source database cataloging AI model specifications, pricing, and capabilities for comparison.
5/22/2026
On AI
Personal perspective on using Claude Code as development tool and importance of code review practices for effectiveness.
5/22/2026
Show HN: Canonry tracks how AI cites you – agent-first, open source
Open-source agent-first tool tracking how AI systems cite and attribute sources.
5/22/2026
Semantic Search Starts With Embeddings
Technical overview of embeddings as foundation for semantic search systems.
5/22/2026
Llmff v0.1.2: FFmpeg-Shaped Pipelines for LLM Workflows
Llmff v0.1.2 tool providing FFmpeg-style pipeline abstractions for LLM workflows. Developer tool for LLM applications.
5/22/2026
SteelSpine: Replay tool for debugging AI agents
Debugging tool for AI agents with event capture, replay, and cryptographic verification. EU AI Act compliant.
5/22/2026
Domain-Camouflaged Injection Attacks Evade Detection in Multi-Agent LLM Systems
Research on domain-camouflaged injection attacks targeting multi-agent LLM systems. Security analysis of agent vulnerabilities.
5/22/2026
OpenCode and Cursor's Composer 2.5
OpenCode project: OpenAI-compatible proxy API for Cursor IDE, enabling standard SDK usage with Cursor's inference engine.
5/22/2026
Frontier labs don't use most AI compute (yet)
Analysis of frontier AI lab compute utilization trends, showing most available compute capacity remains unused despite massive investment.
5/22/2026
Google API keys keep working after you delete them
Security vulnerability: Google API keys remain functional ~23 minutes after deletion, allowing unauthorized access to Gemini and other APIs.
5/22/2026
Gemini Omni
Google announces Gemini Omni Flash: multimodal model generating video, images, and other content from any input type.
5/22/2026
Show HN: CoreMem – Portable context for AI agents
CoreMem: portable context management tool for AI agents. Share context via URL, Chrome extension, MCP, Cursor/VS Code plugins, enabling persistent memory across agent sessions.
5/22/2026
Microsoft Drops Claude Code After Budget Overrun
Brief report: Microsoft cancels internal Claude licenses after unexpected token-based billing costs. Minimal details provided.
5/22/2026
AI Dev 26 x SF | Andi Partovi: Why Every Agent Needs a Simulation Sandbox
Talk on using simulation sandboxes for safe AI agent development and testing.
5/22/2026
AI's Plummeting Prices Are a Software Story, Not a Hardware One
Analysis of why AI inference costs have dropped 70-90% annually, driven by software optimizations rather than hardware, enabling local model deployment on commodity hardware.
5/22/2026
The Invisible Cliff: AI Development and Architectural Debt
Technical analysis of architectural debt patterns in AI systems: feature shipping stability followed by sudden cascading failures requiring rewrites.
5/22/2026
Is Machine Learning still worth learning in 2026?
Career advice on whether machine learning is still worth learning given LLM advances. Opinion piece from ML book author.
5/22/2026
AI Dev 26 x SF | Luke Kim: The Agent Data Stack—Why Every AI Agent Needs Its Own Data Stack
Talk on specialized data infrastructure requirements for AI agents in production.
5/22/2026
How do AI chips work? [video]
Video content explaining how AI chips function at a hardware level.
5/22/2026
AI Dev 26 x SF | Daniel Beutel: Flower SuperGrid Agents
Talk on Flower SuperGrid Agents framework for distributed agent systems.
5/22/2026
AI Dev 26 x SF | Or Dagan: Optimizing Accuracy, Cost, and Latency in Real-World Agents
Talk on optimizing accuracy, cost, and latency tradeoffs for real-world AI agents.
5/22/2026
State of AI 2026
Developer survey of 7,258 respondents on AI's impact on software development work, with 2026 findings on trends and adoption patterns.
5/22/2026
Ad Infinitum
Google I/O announcement of generative UI in Search with AI agents and natural language mini-apps.
5/22/2026
WebGPU back end in llama.cpp/ggml
WebGPU backend implementation added to llama.cpp/ggml for LLM inference acceleration in web environments.
5/22/2026
Agents League: The Esports-Inspired Hackathon Where AI Agents Battle for Glory
Agents League: week-long esports-style hackathon for AI agent development with live coding battles and competitions.
5/22/2026
Show HN: Interactive first-principles climate physics simulation with explainer
Interactive 3D browser-based climate physics simulator built with AI coding assistance (Claude Opus); explains climate processes.
5/22/2026
Show HN: Lilo – An open source personal AI assistant that lives in Telegram
Open source Telegram-based personal AI agent with memory, skills, tools, and persistent workspace; model-agnostic.
5/22/2026
Show HN: OpenRDMA-An Open Source FPGA-Based 400G RDMA-Like SmartNIC
Open-source FPGA-based 400G SmartNIC with LLM-generated testbenches for verification. Uses AI for cocotb Python workflows.
5/22/2026
Fakellm – a mock OpenAI/Anthropic server for testing
Mock OpenAI/Anthropic API server for testing LLM applications and agent flows without API costs.
5/22/2026
AI Dev 26 x SF | Diamond Bishop: The Next 100 Agents. Building the Agent Native Office
Conference talk on building agent-native office software with multiple agents.
5/22/2026
Brethof-mind – local long-term memory for Claude Code (SurrealDB, no API)
MCP server adding persistent, searchable long-term memory to Claude Code using local SurrealDB without APIs.
5/22/2026
AI Dev 26 x SF | Paul Everitt: The Shift to Agentic Engineering
Conference talk on agentic engineering paradigm shift.
5/22/2026
AI Coding Assistants
Personal reflection on using AI coding assistants in development workflow and cognitive impact on learning.
5/22/2026
Use Boring Languages with LLMs
LLMs perform better with consistent, less fragmented programming languages and ecosystems; consistency compounds agentic output quality.
5/22/2026
Launch HN: Superset (YC P26) – IDE for the agents era
Open-source IDE for running multiple coding agents (Claude Code, Codex, OpenCode) in parallel. YC P26 company.
5/22/2026
Show HN: Prisma Next – data contracts, migration graphs, agent DX
TypeScript rewrite of Prisma ORM with data contracts, migration graphs, and integration skills for AI agent tooling.
5/22/2026
Show HN: Blacknode – Visual workflow builder Claude can drive via MCP
Visual workflow builder that Claude can control via MCP (Model Context Protocol) for agent automation.