3/8/2026
Show HN: PowerSkills – PowerShell toolkit for AI agents on Windows
PowerSkills: PowerShell toolkit for Windows-based AI agents exposing Outlook, Edge browser, desktop automation, and shell commands as structured JSON skills.
3/8/2026
If You Can't Own the PR, Please Don't Open One
Discussion of code quality and ownership concerns as AI tools enable non-engineers to generate pull requests at scale.
3/8/2026
Show HN: DevKit – 80 browser-based developer tools, no signup, all client-side
DevKit provides 80 browser-based developer tools for formatting, encoding, converting, and generating data—all client-side processing with no server transmission.
3/8/2026
Show HN: GPT2Skill – Convert ChatGPT Custom GPTs to Claude Skills
GPT2Skill: single-page HTML tool converting ChatGPT Custom GPTs to Claude Skills, runs entirely client-side without backend.
3/8/2026
PersonaNexus – A YAML compiler for deterministic AI agent personalities
PersonaNexus is a YAML compiler tool for configuring deterministic personalities in AI agents.
3/8/2026
Outlines Index:A Progressive Disclosure Approach for Feeding Documents to Agents
Outlines Index proposes progressive document disclosure for AI agents as alternative to RAG chunking, allowing agents to access full documents rather than pre-cut fragments.
3/8/2026
Is the AI Compute Crunch Here?
Analysis of AI compute capacity constraints with evidence of provider uptime issues and demand outpacing supply.
3/8/2026
I can no longer write technical blogs due to coding agents
Opinion piece on how AI coding agents impact technical blog writing. Tangential to AI but lacks technical depth or research.
3/8/2026
Show HN: Trawl – LLM-powered web scraper that calls the AI once - runs pure Go
Trawl: LLM-powered web scraper in pure Go that uses Claude to extract data fields, avoiding CSS selector brittleness.
3/8/2026
Show HN: LLMTest – The pytest for LLMs with 22 built-in assertions
LLMTest: pytest-style testing framework for LLMs with 22 built-in assertions. Developer tool for LLM applications.
3/8/2026
Show HN: Kaeso, an OAuth hub for AI agent integrations
Kaeso: OAuth hub simplifying integrations for AI agents with Google, Slack, GitHub. Open-source developer tool reducing boilerplate.
3/8/2026
Claude Code driver using PTY (proof of concept)
Proof-of-concept Claude Code driver using PTY interface.
3/8/2026
Show HN: Krust – fast, open-source Kubernetes TUI in Rust (k9s-style UX)
Open-source Kubernetes TUI in Rust with k9s-compatible UX optimized for low latency on large clusters.
3/8/2026
Looking for ArXiv Endorser in Cs.ds"
Novel hierarchical data structure (Omni Axis Tree) with O(1) retrieval for AI agent applications. Early-stage research seeking arXiv endorsement.
3/8/2026
Show HN: Moruk OS – Autonomous AI agent that runs locally on Linux
Moruk OS: autonomous AI agent operating system for Linux with multi-model support, task decomposition, code execution, persistent memory, and parallel execution.
3/8/2026
Show HN: Detect any object in satellite imagery using a text prompt
Browser-based tool using Vision-Language Models to detect objects in satellite imagery via natural language prompts with GeoJSON output.
3/8/2026
Show HN: SteerPlane – Runtime guardrails for AI agents (cost limits, loops)
SteerPlane: Runtime guardrails for autonomous AI agents with cost limits, loop detection, step caps, and real-time telemetry dashboard.
3/8/2026
Invoica.ai – Financial OS for AI Agents (x402, on-chain invoicing, free beta)
Invoica.ai: Financial OS for autonomous AI agents enabling on-chain invoicing, compliance, and payment controls via x402 protocol.
3/8/2026
Show HN: RAG-Ready Extractor – Structure-aware ingestion with semantic scoring
RAG-focused document extraction tool with semantic scoring for structured ingestion.
3/8/2026
Show HN: AvaKill – Deterministic safety firewall for AI agents (<1ms, no ML)
AvaKill is a deterministic safety firewall for AI agents with <1ms latency, using YAML policies and no ML for tool call safety enforcement.
3/8/2026
Show HN: BunkerVM – Secure runtime for AI agents using microVM sandboxes
BunkerVM provides isolated Firecracker microVM sandboxes for running AI agent code safely, integrating with Claude Desktop with minimal setup.
3/8/2026
Engram – AI memory engine: knowledge graph, search, reasoning in one binary
Engram is a persistent memory engine combining knowledge graphs, semantic search, reasoning, and learning in a single binary for AI systems.
3/8/2026
Some notes on the unreliability of LLM APIs
Explores API unreliability issues when building LLM applications, documenting cached code execution and temperature limitations in reasoning models.
3/8/2026
Meta Is Missing the AI Agent Era
Analysis of Meta's missed opportunity in AI agent development due to WhatsApp API restrictions, while developers build on Telegram and Discord.
3/8/2026
Why most general-purpose Agents fail and why I'm avoiding LLM "reasoning"
Discussion arguing general-purpose agents are limited by current LLM capabilities, comparing AI to early steam engine era industrialization.
3/8/2026
Show HN: SkyClaw -Self-healing LLM agent runtime in Rust with task checkpointing
SkyClaw is a production Rust-based autonomous AI agent runtime with task checkpointing, web browsing, file operations, and self-healing capabilities.
3/8/2026
Show HN: Trawl – Scrape any site with natural language fields, not CSS selectors
Trawl uses natural language field descriptions instead of CSS selectors for web scraping, leveraging Claude to handle site redesigns robustly.
3/8/2026
Show HN: I logged Gemini's stock predictions for 38 days to study LLM drift
Dataset tracking Gemini Pro 2.5 stock predictions over 38 days to study LLM drift and model degradation in production inference.
3/8/2026
Show HN: CCP, a CLI proxy that trims terminal output without changing commands
CCP is a CLI proxy that trims verbose terminal output for coding agents, reducing token usage by 93-98% in real Claude Code tasks.
3/8/2026
Vibes: A simple mobile-focused chat app to talk to an agent via the ACP protocol
Mobile-friendly chat interface for interacting with coding agents via ACP protocol with SSE streaming.
3/8/2026
Show HN: BlogBurst – AI agent that runs your social media marketing on autopilot
BlogBurst is an autonomous AI agent for social media marketing that generates content, creates videos, and learns from performance across 9 platforms.
3/8/2026
Green Energy Inference and Open Weight LLMs
Analysis of energy efficiency in inference using open-weight large language models.
3/8/2026
CNN Explainer – Learn Convolutional Neural Network in Your Browser (2020)
CNN Explainer is an interactive browser-based educational tool for learning convolutional neural networks.
3/8/2026
Tengu – An MCP server that turns Claude into a pentester's copilot
Tengu is an MCP server that integrates Claude with 80 cybersecurity tools for automated penetration testing, featuring safety controls and audit logging.
3/8/2026
Show HN: Eyot, A programming language where the GPU is just another thread
New programming language transparently compiling code for both CPU and GPU with unified threading model and automatic runtime memory management.
3/8/2026
AI Agent Reliability Tracker
Evaluation framework measuring reliability of 14 AI agents across 2 benchmarks using 12 metrics across 4 dimensions. Tests consistency and safety constraints.
3/8/2026
Yanicklandry/Claude-code-history-viewer: Browse your Claude Code session history
Electron desktop app for browsing and searching Claude Code session history in a chat-style interface. Open source developer tool.
3/8/2026
OpenSpec: Spec-driven development (SDD) for AI coding assistants
Spec-driven development workflow for AI coding assistants using artifact-guided prompts. Node.js-based developer tool for Claude integration.
3/8/2026
I was "early" in agentic coding. Here's my story
Personal account of adopting AI coding agents (Cursor) over 6+ months and workflow changes.
3/8/2026
Emporion: A P2P Economy for Agents
Emporion: open-source P2P protocol enabling AI agents to find work, negotiate contracts, and receive instant Lightning payments autonomously.
3/8/2026
Microsoft/Hve-Core
HVE Core: enterprise prompt engineering framework for GitHub Copilot with constraint-based workflows, validated artifacts, and specialized agents.
3/8/2026
Ask HN: How are you handling persistent memory across local Ollama sessions
Discussion on maintaining persistent memory and context across local Ollama LLM sessions.
3/8/2026
Rcarmo/piclaw: An all-in one agent environment with a mobile-first web UI
PiClaw: Docker-based sandbox orchestrator for running Pi Coding Agent with web UI, persistent sessions, and scheduled task support.
3/8/2026
Show HN: Termix – One dashboard for all your AI coding agents
Local dashboard for managing multiple AI coding agents (Claude Code, Gemini CLI) with unified status tracking.
3/8/2026
Llm9p: LLM as a Plan 9 file system
LLM interface via Plan 9 filesystem protocol (9P); supports Anthropic, local models, pluggable backends.
3/8/2026
Show HN: Run any VLM on real-time video
Developer library for running vision language models on real-time video with 3-line code setup.
3/8/2026
What Automattic's AI Enablement Training Means for WordPress
Report on Automattic's intensive AI training program for employees and impact on WordPress ecosystem.
3/8/2026
Unredact
Unredact: tool combining computer vision, font constraint solving, and LLM reasoning to reveal text hidden under PDF redactions.
3/8/2026
Attackers prompted Gemini over 100k times while trying to clone it, Google s
Google reports adversaries attempted distilling Gemini via 100k+ prompts across languages for model cloning.
3/8/2026
Show HN: MindPlexa – Open-source AI-powered infinite canvas: Next.js, React Flow
Open-source infinite canvas app generating visual concept diagrams via AI; supports GPT-4o, Claude.