3/8/2026
Aiswitch – switch between Claude, OpenAI, Gemini and Copilot accounts in one cmd
Aiswitch: CLI tool to switch between Claude, OpenAI, Gemini, and Copilot accounts for developers using multiple AI services simultaneously.
3/8/2026
FastFlowLM Docker – Run LLMs on AMD Ryzen AI NPU (Linux)
Docker setup to run LLMs on AMD Ryzen AI NPU under Linux using FastFlowLM, generated by Claude Opus with human guidance via Copilot CLI.
3/8/2026
Show HN: Agentlytics – See stats from all your AI coding agents (100% local)
Local analytics tool aggregating statistics from multiple AI coding agents (Cursor, Windsurf, Claude Code, Copilot) including costs, token usage, and session tracking across 14 editors.
3/8/2026
Show HN: Admiran: a pure, lazy functional language and self-hosting compiler
Self-hosting pure lazy functional language compiler (~7kloc) in itself, implementing pattern matching, algebraic data types, and functional data structures for educational exploration.
3/8/2026
Ask HN: How are you monitoring AI agents in production?
Discussion of production AI agent monitoring, covering failure modes like untracked token usage, undetectable outputs, and lack of audit trails. References AgentShield observability tool.
3/8/2026
Show HN: I built a pipeline that generates a comedy podcast end-to-end with AI
Automated end-to-end comedy podcast generation pipeline using Gemini, Temporal orchestration, and ElevenLabs voice synthesis.
3/8/2026
Agent VM – Lima VMs for local agents and more
Sandboxed VM environment using Lima to safely run AI coding agents (Claude Code, OpenCode) with full autonomy while isolating host system, ships with Docker and headless Chrome.
3/8/2026
Show HN: Promptinel – A Security Scanner for Prompts
Security scanner tool for prompts that identifies vulnerabilities before runtime, treating prompts as executable artifacts requiring security analysis.
3/8/2026
Show HN: Open-source alternative to Wispr Flow
Open-source voice-to-text tool using Whisper for transcription and LLM for output formatting. Hotkey-activated, clipboard-integrated alternative to Wispr Flow.
3/8/2026
Could ternary weights make 500B models runnable on consumer hardware?
Open-source inference engine using ternary quantization (1.58-bit) to compress models 10x, enabling 500B+ parameter models on consumer hardware via INT8/INT4 quantization, speculative decoding, and dynamic sparsity.
3/8/2026
The Problem with AI Agents Isn't Identity, It's Authorization
Analysis of AI agent security challenges, arguing authorization scoping is more critical than identity verification. Discusses granular API access control for agent containment.
3/8/2026
How Gen AI Is Changing the Way We Write Code
Opinion on how LLMs like GPT/Claude changed software development workflow from methodical to collaborative iteration.
3/8/2026
Show HN: I built a tool to search 50K+ AI coding conversations in SQLite
SQLite tool for searching 50K+ AI coding conversations from Claude/Codex. Full-text search with incremental sync.
3/8/2026
Show HN: Open-source AI SaaS to turn long videos into viral shorts
Open-source tool using Whisper and Gemini 2.0 Flash to auto-generate viral short clips from long videos.
3/8/2026
Show HN: I built a local AI crypto monitoring agent that texts me on Telegram
AI agent monitoring cryptocurrency markets and portfolios, sends Telegram alerts. Integrates multiple data sources.
3/8/2026
Show HN: LLM hallucinates its carbon footprint, we measure and plot it on steam [video]
Art installation using local LLM to generate SVG while measuring real power consumption, visualizing hallucinations on steam-covered glass.
3/8/2026
Beachpatrol: Browser's End-User Automation CLI Hub
CLI tool to automate web browsers via Playwright scripts for end-user automation workflows.
3/8/2026
Show HN: I built AgentSafety, an open benchmark for coding-agent safety
Open benchmark testing autonomous coding agents on safety decisions: allow, ask, refuse. 50 test cases covering prompt injection, secrets, destructive commands.
3/8/2026
Show HN: UltraContext – open-source shared context layer for coding agents
Open-source context layer for coding agents that captures and shares session data between agents in real-time, enabling better coordination and information transfer.
3/8/2026
The Shift from ML Engineering to AI Engineering
Book review series on AI Engineering by Chip Huyen, covering differences between ML and AI engineering, foundation models, and building applications on top of them.
3/8/2026
AI doesn't replace white collar work
Opinion arguing AI doesn't replace white-collar knowledge work, emphasizing human judgment value.
3/8/2026
Show HN: Skales – Local AI agent desktop app (.exe/.dmg, 300MB idle RAM)
Desktop AI agent application (.exe/.dmg) designed for low resource usage (300MB idle RAM), making AI agents accessible to non-technical users.
3/8/2026
Commitaai – An open-source CLI to generate conventional commits using AI
Open-source CLI tool using AI to automatically generate conventional commit messages from staged Git changes, improving code organization.
3/8/2026
Show HN: I over-engineered a home security camera that uses an LLM and talks
Open-source Python pipeline for home security camera using local Vision LLM and text-to-speech for scene analysis.
3/8/2026
Show HN: Claude Code skill that generates ship pages from one sentence
Claude Code Skill that generates production-ready landing pages from single-sentence prompts using LLM.
3/8/2026
Your Agent Doesn't Need a Readme
Article arguing AI agents need structured data formats instead of README files for effective navigation and tool usage.
3/8/2026
Show HN: Claude Skill for temporary cost tracking
Claude Skill for temporary cost tracking during interactive Claude Code sessions to monitor API token usage and expenses.
3/8/2026
Show HN: GCI, Fly.io like deployments with any VM
CLI deployment tool using SSH and Docker Swarm for VMs as alternative to Fly.io addressing cold starts and costs.
3/8/2026
Show HN: Svitrix – open-source firmware for Ulanzi TC001 pixel clock
Open-source ESP32 firmware for Ulanzi pixel clocks with bug fixes, tests, Home Assistant integration, and browser flasher.
3/8/2026
Show HN: I open-sourced the AI agents I use to run 35 marketing clients solo
Open-source repository of AI agents for marketing automation covering ad management, website building, and SEO for 35+ clients.
3/8/2026
Show HN: Think Better – 155 decision-science rules for your AI assistant
Think Better: open-source AI skills package with 155 decision-science frameworks for LLM assistants, following Anthropic skill format.
3/8/2026
Show HN: System R – Risk intelligence API for autonomous trading agents
System R provides risk management API for autonomous trading agents with examples in Python, CrewAI, and LangChain. Blocked 53 trades, reduced max drawdown 29 percentage points.
3/8/2026
Show HN: I made Qwen3.5-4B 13% smarter by compressing it to 4-bit
Mixed-precision 4-bit quantization technique improves Qwen3.5-4B model performance, achieving lower perplexity than FP16 baseline while reducing VRAM usage to 6.4GB.
3/8/2026
Why Task Proficiency Doesn't Equal AI Autonomy
Critical examination of claims that AI agents will replace software engineers, analyzing autonomy vs task proficiency gap.
3/8/2026
Show HN: Lightweight agent proxy for monitoring sensitive data exposure
Go CLI proxy for monitoring LLM API requests and flagging sensitive data exposure in outbound agent traffic.
3/8/2026
The Architecture Review I'd Give Your Multi-Agent System
Architectural patterns for multi-agent systems addressing six common mistakes: inter-agent coordination, task management, scalability, cost optimization.
3/8/2026
Autonomous AI code conversion experiment
Python implementation of LCModel converted from Fortran using OpenAI Codex with project-owner oversight and validation.
3/8/2026
I Built a 2300-File Research Workspace with AI: Here's What It Does
NimoNova: AI-assisted research workspace that organizes saved articles, papers, PDFs, and images into a knowledge system. Part 4 of series on AI-assisted development.
3/8/2026
Cloud Sandbox Benchmarks
Benchmark comparison of cloud sandbox providers (Daytona, E2B, Blaxel, Modal) for Python code execution performance.
3/8/2026
RailsForge – a Rails development toolkit I built with AI
RailsForge: CLI tool for auto-generating Rails components including monitoring, DevOps, security analyzers, and Docker/CI-CD configurations.
3/8/2026
Orchestration for zero-human companies
Open-source platform to orchestrate AI agents as employees with role-based assignments, budgets, and monitoring dashboard.
3/8/2026
AI Agents Have an Information Cocoon Problem No One Talks About
Analysis of information cocoon problem in AI agents: agents stuck in loops, duplicating code, failing to search existing solutions.
3/8/2026
Show HN: Harden – 5 AI models audit your code, then debate each other's findings
Developer tool running 5 frontier models in parallel for code audit, models debate findings, coordinator synthesizes consensus. Claude, GPT-4o, Gemini, Mistral, DeepSeek.
3/8/2026
Show HN: AngelLab – modular Linux host monitoring with autonomous agents
Open-source modular Linux host monitoring system using autonomous agent workers for subsystem surveillance. Structured event aggregation with CLI and Prometheus exposure.
3/8/2026
Show HN: Open-source Cursor cost tracker with anomaly detection and Slack alerts
Open-source tool tracking Cursor IDE costs with anomaly detection and Slack alerting. Developer tool for LLM application costs.
3/8/2026
Show HN: Agentcheck – Check what an AI agent can access before you run it
Open-source CLI tool scanning shell for credentials and access that AI agents could exploit. Covers AWS, GCP, Azure, API keys, K8s, Docker.
3/8/2026
Show HN: I'm building an open source alternative to Topaz Photo AI
Open source alternative to Topaz Photo AI with direct implementation of all AI logic (preprocessing, inference, tiling, tensor ops) without external APIs.
3/8/2026
Show HN: Claude Code Container – Zero-Config Docker Isolation for Claude Code
Claude Code Container: zero-config Docker isolation for Claude's code generation, balancing safety and developer flow without permission prompts.
3/8/2026
Show HN: LLMTor – Access Public LLMs anonymously using blind signatures and Tor
LLMTor: tool enabling anonymous access to public LLMs using blind signatures and Tor, preventing server operators from linking prompts to users.
3/8/2026
Show HN: Atombot – tiny personal assistant for local models and GPT‑5.4
Atombot: open source self-hosted personal AI assistant (~500 lines) with persistent memory, Telegram gateway, reminders, and extensible skills system for local and GPT models.