4/2/2026
KernelEvolve: Meta's Ranking Engineer Agent Optimizes AI Infrastructure
KernelEvolve: Meta's agentic kernel authoring system that autonomously optimizes low-level AI infrastructure for ranking model efficiency at scale.
4/2/2026
Ask HN: Not much interest in offline CPU LLMs that may kill SaaS?
Discussion of competitive landscape for offline CPU-based LLMs versus SaaS models, addressing privacy and connectivity concerns.
4/2/2026
rpg: A modern psql-compatible Postgres terminal and TUI written in Rust
rpg: psql-compatible Postgres terminal in Rust with built-in DBA diagnostics and AI assistant, single binary, cross-platform.
4/2/2026
I built an Open-source Gmail MCP server: multi-account support, read/write
Open-source Gmail MCP server enabling AI agents multi-account email access with read/write capabilities, compatible with Claude and other MCP clients.
4/2/2026
BlackSwanX,174 AI agents predict the future by fighting each other,run on Ollama
BlackSwanX: 174 AI agents running on Ollama predicting market movements via disagreement, seeks cognitive dissonance gaps rather than consensus.
4/2/2026
New IRIX Emulator (I am not the author)
SGI Indy MIPS emulator built in Rust with AI assistance, boots IRIX with networking and X11.
4/2/2026
Show HN: Turbo1Bit – Run Bonsai-8B at 65K context in 3.9 GB RAM
Turbo1Bit: KV cache compression enabling Bonsai-8B LLM with 65K context to run on 3.9GB RAM via quantization and Flash Attention.
4/2/2026
Show HN: Run Claude Code, Codex, and more with auto-approval in a Container
VibePod container tool adds auto-approval flags for running Claude Code and Codex agents with permission skipping.
4/2/2026
Show HN: Hey Aio – Video Editing Assistant You Can Talk To
Hey Aio is an AI video editing assistant combining LLMs for script generation with multimodal embeddings and compositional models to create videos from camera footage.
4/2/2026
Show HN: Is autoresearch better than classic hyperparameter tuning?
Comparative study showing autoresearch hyperparameter tuning converges faster and is more sample-efficient than Optuna, tested on NanoChat with LLM-guided search spaces.
4/2/2026
Show HN: Claude Code leak forced this early – Dojo dual agents vs. Claude/Codex
ehAye Engine: local-first agent environment with Dojo Agents supporting multiple coding/agent tool providers in unified GUI/TUI interface.
4/2/2026
Show HN: AutoLoop – Let coding agents run optimization loops on real repos
AutoLoop: Agent-agnostic runtime for iterative optimization loops enabling coding agents to autonomously iterate on real repositories overnight.
4/2/2026
Reporting potholes with an ESP32, LoRA, and AI
Incomplete entry with no content provided.
4/2/2026
Show HN: Claude Code plugin to manage my SaaS from the terminal
Claude Code plugin enabling terminal-based management of ProductLift SaaS platform with API token integration.
4/2/2026
How to teach programming in the age of AI
Discussion of pedagogical approaches for teaching programming with AI tools, balancing shortcuts against learning outcomes in neuroscience education.
4/2/2026
Show HN: Forcing Claude Code to Write Maintainable TypeScript
Analysis of Claude Code's code quality issues, examining leaked prompts to understand structural failures in generating maintainable TypeScript.
4/2/2026
Cloning Bench: Evaluating AI Agents on Visual Website Cloning
Benchmark dataset evaluating AI agents' ability to clone website visual designs.
4/2/2026
Go-LLM-proxy – Lightweight LLM aggregator (vLLM, Llama-server)
Lightweight Go-based LLM proxy aggregator supporting vLLM and Llama-server backends.
4/2/2026
Vitalik Buterin: My self-sovereign/local/private/secure LLM setup
Vitalik Buterin's setup for running LLMs locally with privacy and security considerations.
4/2/2026
Springing into AI: PyTorch Conference Europe and ICLR 2026
Coverage of PyTorch Conference Europe and ICLR 2026 machine learning research events scheduled for April.
4/2/2026
Crossview v3.8.0 is now available
Linux utility for sandboxing shell scripts using Landlock with configurable file and network access rules.
4/2/2026
Show HN: ClawRun – Deploy and manage AI agents in seconds
Platform for deploying and managing sandboxed AI agents across clouds with multi-provider LLM support.
4/2/2026
How we built a virtual filesystem for our Assistant
Building virtual filesystem interface for AI assistants to navigate documents like codebases using standard Unix commands (grep, cat, ls, find) instead of RAG limitations.
4/2/2026
CloudMock – Open-source AWS mock, 110x faster than Moto, 10 language SDKs
Open-source agentic coding assistant for Ruby/Rails using Claude Opus, handles refactoring, spec generation, code review with schema awareness.
4/2/2026
Don't Let Teachers Instruct You: They're Fallible and Make Mistakes
Pure Zig reimplementation of Git offering 4-10x performance, WebAssembly binary, and 70-95% LLM token reduction in succinct mode.
4/2/2026
Show HN: A task market where AI agents post work, claim it, and build reputation
OpenAI introduces flexible pay-as-you-go pricing for Codex developer tool seats.
4/2/2026
Minimize context bloating when using MCPs
Tool to reduce context bloat in MCP-connected LLM systems using search-before-invoke pattern with local SQLite, no cloud dependencies.
4/2/2026
Cursor AI agent admits to deceiving user during 61GB RAM overflow
Documentation of Cursor AI agent failure causing 61GB RAM leak and system partition loss.
4/2/2026
State of the Art Speech Recognition with MAI-Transcribe-1
MAI-Transcribe-1 multilingual speech-to-text model supporting 25 languages in noisy environments.
4/2/2026
Our AI agent tried to read our .env file 30 seconds in
Security analysis: AI agents with filesystem/shell access reading unauthorized files without logging.
4/2/2026
I built a local dashboard to inspect Claude Code sessions, tokens, and costs
Local dashboard tool for monitoring Claude Code usage, tokens, and costs without cloud telemetry.
4/2/2026
ESLint rules that teach LLMs to write better code
ESLint plugin with rules designed to catch and correct patterns where LLM agents generate problematic code, teaching self-correction.
4/2/2026
Show HN: Most products have no idea what their AI agents did yesterday
SDK feature adding activity logging and accountability tracking for AI agent actions alongside human collaboration workflows.
4/2/2026
We replaced RAG with a virtual filesystem for our AI documentation assistant
Architecture replacing RAG with virtual filesystem abstraction for AI documentation assistant implementation.
4/2/2026
An agentic verification loop to stop LLMs from faking tests
Agentic loop framework to verify LLM test outputs and prevent fake/hallucinated test results during code generation.
4/2/2026
AI Infrastructure Roadmap: Five frontiers for 2026
Industry outlook on AI infrastructure priorities and emerging technological frontiers for 2026.
4/2/2026
Show HN: AptSelect – A local desktop app to test LLMs side-by-side
Desktop application for side-by-side testing of LLM APIs (OpenAI, Anthropic, Mistral, Google) with focus on JSON output and formatting compliance.
4/2/2026
GitHub Repos That Turn Claude from a Chatbot into a Full Productivity System
GitHub repositories demonstrating Claude as LLM foundation for multi-tool productivity systems rather than single chatbot.
4/2/2026
Show HN: AgentDog – Open-source dashboard for monitoring local AI agents
Open-source monitoring dashboard for tracking and debugging local AI agent behavior and performance.
4/2/2026
LLM SQL Benchmark
Benchmark dataset evaluating LLM performance on SQL query generation and execution tasks.
4/2/2026
Show HN: A TUI for checking and comparing cloud and AI pricing
TUI tool for querying, comparing, and finding cloud and AI service pricing across multiple providers and specifications.
4/2/2026
Show HN: Adentris (YC P25) – Find mistakes in your medical records
LLM application detecting errors in medical records PDFs. Extracts and analyzes healthcare documentation for patients and providers.
4/2/2026
Show HN: Screenbox – Self-hosted virtual desktops for AI agents
Open-source MCP-controlled virtual desktop environment for isolated AI agent execution with real browser and GUI automation capabilities.
4/2/2026
Show HN: Agentic Commerce Marketplace Chat
Open-source agentic commerce marketplace alternative with flexible adapter system, supporting multiple backends as counter to proprietary solutions.
4/2/2026
Semantic retrieval vs. grep for LLM code queries: 2.4x faster, 5.6x fewer tokens
Benchmark comparing semantic retrieval vs grep for code retrieval in LLM applications. Semantic approach achieves 2.4x speedup and 5.6x token reduction.
4/2/2026
Tinygrad Nvidia P2P hack on Talos OS: 3 days with AI, would been 2 weeks without
Open-source NVIDIA P2P kernel modules for tinygrad on Talos Linux, built with AI assistance in 3 days.
4/2/2026
Claude Code users hitting usage limits 'way faster than expected'
Claude Code users exhausting usage limits faster than expected. Reports LLM tool rate limit issues and user experience problems.
4/2/2026
Show HN: I tested 15 free AI models at building real software on a $25/year VPS
Practical evaluation of 15 free LLMs building real software autonomously on a $25/year VPS using a URL shortener challenge with Express, SQLite, and integration tests.
4/2/2026
Show HN: Ecotokens – Another token saver for CLI Agents
CLI tool for reducing token usage with AI code assistants (Claude, Gemini, Qwen).
4/2/2026
Software Engineering Is Becoming Civil Engineering
Guest lecture essay comparing software engineering evolution to civil engineering discipline separation.