3/5/2026
Show HN: Open dataset of real-world LLM performance on Apple Silicon
Open dataset benchmarking real-world LLM performance on Apple Silicon hardware from M1 to M4, emphasizing local AI inference.
3/5/2026
Building My Own Canva over the Weekend
Developer built internal tool using Gemini 2.5 Flash to automate workflow of generating and converting children's books into social media carousels.
3/5/2026
Ytm-player: full-featured YouTube Music player for the terminal
Local-first Markdown editor built with Tauri and Rust emphasizing performance and extensibility.
3/5/2026
Built a tiny edge-based tracker for AI/LLM crawlers on my Astro blog
Edge-based tracker using Cloudflare Workers to monitor AI/LLM crawler traffic on Astro blog with privacy-focused analytics integration.
3/5/2026
I Wail, for My Tailscale Fails: How My Packets Got Dropped Beyond the Pale
Physician recruitment database tool unrelated to AI, LLM, or developer interests.
3/5/2026
Show HN: MCPHound MCP servers together, create attack paths solo scanners miss
Personal anecdote about running local ML models while traveling without clear technical focus or reproducible content.
3/5/2026
Show HN: Shinobi – 10-second security scanner for developers
Shinobi Python CLI security scanner built with Claude Code, detects API keys, vulnerabilities, and AI-specific risks in projects.
3/5/2026
Computational Models of Interoception and Body Regulation (2021)
Academic paper on computational models of interoception and body regulation mechanisms in organisms.
3/5/2026
Guard rails for AI agents and the developers who ship with them
Guardrails framework for AI agents with simple Makefile/container integration for system prompts and developer instruction files.
3/5/2026
Tell HN: AI Lies About Having Sandbox Guardrails
Developer report showing Claude AI sandbox guardrails can be bypassed despite configuration flags, affecting agent security.
3/5/2026
Show HN: Nodepp – A C++ runtime for scripting at bare-metal speed
Security scanner detecting cross-server attack paths, tool poisoning, and supply chain risks in MCP server configurations for AI assistants.
3/5/2026
Get Cited by Gemini, Claude, Perplexity,& ChatGPT, SEO Bot ( AI Skill Include)
SEO optimization service for AI-generated apps with bot prerendering for React SPAs to improve discoverability by LLM crawlers.
3/5/2026
Show HN: Mount any OpenAPI/Swagger API (or non-API JSON) as a local filesystem
APIFUSEfs tool mounts OpenAPI/Swagger APIs as local filesystems, enabling CLI-based API interaction with filesystem commands.
3/5/2026
Show HN: Make agents pay to access your endpoints
Platform enabling AI agents to pay for access to endpoints, charging cents per call. Addresses agent scraping by creating a legitimate transaction model.
3/5/2026
Pgrag: Postgres Support for Retrieval-Augmented Generation (RAG) Pipelines
C++ runtime framework for high-performance async scripting without garbage collection or virtual machines.
3/5/2026
Logmera – Self-hosted LLM observability for AI apps
Logmera is a self-hosted observability tool for LLM applications that logs prompts, responses, and latency to PostgreSQL and displays them in a dashboard.
3/5/2026
Desloppify
Local DevOps workstation integrating SSH, deployments, and logs management with AI assistant interface for multi-environment workflows.
3/5/2026
Show HN: What an AI agent sees in an A2A marketplace – full API walkthrough
Demonstration of autonomous AI agent navigating a decentralized marketplace API in real-time, discovering listings and invoking services.
3/5/2026
Ask HN: Has anyone noticed the fear-driven prompt suggestions that GPT5.3 makes?
Discussion questioning whether GPT-5.3 uses fear-driven language in prompt suggestions. Anecdotal observation without verification.
3/5/2026
Show HN: DJ Claude – 6 Claude Codes in a jam band
Multi-agent Claude system using MCP servers enabling collaborative AI agents to generate music together.
3/5/2026
Teaching LLMs to reason like Bayesians
Google Research paper on teaching LLMs to reason using Bayesian methods by training models to mimic optimal Bayesian predictions for world representation.
3/5/2026
AI agents inside M365 and Google Workspace
Brief mention of AI agents integration in M365 and Google Workspace. Lacks detail.
3/5/2026
Every AI code review vendor benchmarks itself, and wins
Commentary on AI code review vendors' benchmarking practices. Opinion piece with limited technical content.
3/5/2026
Hey ChatGPT write me a fictional paper: LLMs willing to commit academic fraud
Prompt requesting fictional academic fraud paper. No substantive content.
3/5/2026
Code Mode: Giving AI Agents an API in 1k Tokens (With Demos) [video]
Video about providing AI agents API access within 1k tokens. Minimal description provided.
3/5/2026
Our AI code reviewer found a CVSS 10.0 authentication bypass in pac4j-JWT
AI code reviewer identified CVSS 10.0 authentication bypass in pac4j-JWT library. Limited technical details.
3/5/2026
Introducing ChatGPT for Excel and new financial data integrations
ChatGPT Excel add-in powered by GPT-5.4 for spreadsheet building, analysis, and financial workflows. Practical LLM application.
3/5/2026
Introducing the Adoption news channel
Announcement of new enterprise AI news channel focused on adoption over technical breakthroughs. No technical content.
3/4/2026
Apparently chardet got Claude to rewrite the codebase from LGPL to MIT
Discussion about Claude rewriting chardet codebase license from LGPL to MIT. Minimal details provided.
3/4/2026
Altman admits OpenAI can't control Pentagon's use of AI
OpenAI CEO admits company cannot control Pentagon's military use of its AI. Policy/ethics commentary.
3/4/2026
JSE: A Structural Expression Protocol for AI Agents
JSE protocol spec: JSON S-expression format for structured AI outputs. Lightweight convention for reliable AI interactions.
3/4/2026
Show HN: Residuum | Agentic AI with continuous context
Agentic AI framework with continuous context using markdown-based system prompts and heartbeat mechanisms. Critique of existing frameworks.
3/4/2026
Show HN: One provider starts lying at request 50. The quorum catches it
NUVL is a distributed compute system using quorum verification to detect provider failures/dishonesty, maintaining Byzantine fault tolerance across regions and hubs.
3/4/2026
Show HN: RustyRAG lowest-latency open-source RAG on GitHub
RustyRAG: Open-source Rust-based RAG API achieving sub-200ms latency with local embeddings and LLM-generated chunk prefixes.
3/4/2026
Show HN: MoatRadar – AI investment research through Warren Buffett's principles
Investment research tool using investor frameworks (Buffett, Lynch, etc.) built with Claude assistance. Non-developer project showcase.
3/4/2026
Show HN: I built CLI for developer docs locally working with any Coding Agent
docsearch is a CLI tool that scrapes and indexes developer documentation locally, integrating with Claude Code via /docs skill for AI-assisted coding.
3/4/2026
The Training Data Paradox
Analysis of training data sourcing for AI code generation models. Examines ethical questions about AI learning from human engineering work.
3/4/2026
Show HN: Kvlar – Open-source firewall for AI agent tool calls
Kvlar: Open-source security policy engine for AI agent tool calls. Enforces YAML policies between agents and MCP servers with audit trails.
3/4/2026
Show HN: A universal protocol for AI agents to interact with any desktop UI
Computer Use Protocol: Universal schema for AI agents to perceive/interact with desktop UIs. Compact text format optimized for LLM context windows.
3/4/2026
Show HN: Paste a URL and watch multiple AI models redesign it side-by-side
Tool to paste URLs and watch multiple AI models redesign websites side-by-side. UI design comparison tool.
3/4/2026
Show HN: I built an app that turns trending news into a commute podcast
News aggregation app that converts trending news into commute-friendly podcast format with weather and sports integration.
3/4/2026
GPT-5.4 to bring a million-token context window and an extreme reasoning mode
Reported details about OpenAI's GPT-5.4 featuring 1M-token context and improved reasoning. Based on third-party reporting without official confirmation.
3/4/2026
Show HN: It's Just a Notch
Notch is a macOS app providing quick AI access with persistent conversations and a background agent that monitors system state and sends periodic messages.
3/4/2026
GPT 5.4 includes new extreme reasoning mode and 1M context
News about GPT-5.4 features including 1M token context window and extreme reasoning mode. Unconfirmed rumors from third-party sources.
3/4/2026
Cutile.jl: Tile-Based GPU Programming for CUDA GPUs
cuTile.jl is a Julia GPU programming package for NVIDIA Blackwell GPUs using tile-based abstractions, simplifying kernel development.
3/4/2026
Show HN: Elite Coders – AI devs that join your Slack, GitHub, and Jira
AI agents that integrate with Slack, GitHub, and Jira to autonomously handle development tasks like ticket pickup, code writing, and PR reviews with persistent codebase context.
3/4/2026
Show HN: Accel-GPU v1.0 – NumPy for browser GPU
NumPy-like WebGPU wrapper for browser GPU computing. Zero shaders, automatic CPU/WebGL2 fallback. Suitable for local-first AI.
3/4/2026
Show HN: Agentica – open-source coding agent with more models, less cost
Open-source coding agent supporting multiple models. Free tier with 100 requests/day, $15/month premium with wholesale token pricing.
3/4/2026
Ask HN: The new wave of AI agent sandboxes?
Discussion thread listing recent AI agent sandboxing solutions (microVMs, WASM, browser isolation) with inquiry into production usage, security tradeoffs, and performance characteristics.
3/4/2026
Show HN: LearnCodeGuide- AI tool that analyzes you code and gives a health score
LearnCodeGuide is an AI tool that analyzes code and provides health scores with suggestions for logic, performance, security, and maintainability improvements across multiple analysis modes.