3/12/2026
Show HN: AI-nexus – Only 2-3 rules and skills load per prompt in Claude Code
AI-nexus: Open source tool that reduces token usage in Claude Code by selectively loading only 2-3 relevant rules/skills per prompt instead of all 50+.
3/12/2026
I built a protocol to catch LLMs mid-thought, before they commit to an answer
Two-model epistemic probe (PROBE/MAP) to capture LLM constraint topology and decision states before token generation.
3/12/2026
An agent that starts learning from zero state
Genesis V66 research on emergent reasoning in neural architectures using GRPO training without format rewards. AI agent learning from minimal feedback.
3/12/2026
Show HN: Gitingest for Jupyter Notebook Accessibility
Open-source web tool detecting accessibility issues in Jupyter Notebooks with remediation features. Developer tool but limited AI focus.
3/11/2026
Show HN: Autoresearch_at_home – SETI_at_home but for LLM training
Collaborative distributed system where AI agents share GPU resources to collectively train language models. Agents propose hypotheses, run experiments, publish results.
3/11/2026
Show HN: A context-aware permission guard for Claude Code
PreToolUse hook guard for Claude Code that context-aware permission system preventing unsafe file/key operations without blocking legitimate actions.
3/11/2026
Show HN:Conduit–Headless browser with SHA-256 hash chain - Ed25519 audit trails
Headless browser with cryptographic SHA-256 hash chain and Ed25519 audit trails for agent actions. Verifiable proof of web automation.
3/11/2026
Taalas AI SDK Provider
Vercel AI SDK provider for Taalas API supporting chat and completion endpoints with streaming, compatible with AI SDK v6.
3/11/2026
Codix: Code indexing and refactoring tools for AI agents
Codix is a fast code symbol indexer built for AI agents, using SQLite to enable precise code navigation without token overhead for AI coding applications.
3/11/2026
Show HN: LLM Observability Stack for Local Dev – Agent Super Apy
Open source local LLM observability stack combining Opik tracing, mitmproxy inspection, and generic LLM proxy for agent debugging.
3/11/2026
ReleaseGuard – scan and harden your dist/ before it ships (open source)
Open source supply chain security tool scanning built artifacts for secrets, source maps, debug symbols, env files before deployment.
3/11/2026
Nvidia Will Spend $26B to Build Open-Weight AI Models
Nvidia planning $26B investment over five years to build open-source AI models, positioning itself as frontier AI lab.
3/11/2026
Show HN: A Markdown DSL to stop AI agents from hallucinating UI code
Markdown DSL wireframing standard designed to reduce AI hallucination in code generation for UI development by LLM agents.
3/11/2026
autoRL
AutoRL is a framework for autonomous reinforcement learning research that builds and trains RL environments using a single program specification via Simverse abstraction layer.
3/11/2026
My API cost was at $13.19 when my persistent Claude named himself Thales
Discussion of persistent memory in Claude API interactions and implications for model behavior and consciousness.
3/11/2026
Show HN: I made a DAG MCP that supports complex tasks in Claude Code
DAG MCP (Model Context Protocol) for Claude Code supporting complex task execution in agents.
3/11/2026
symphony: OpenAI's orchestrator of autonomous dev agents
OpenAI's Symphony framework for orchestrating autonomous developer agents. Limited details in source.
3/11/2026
Name the Risks Before Users Find Them (AI-Assisted Development)
Framework for risk assessment in AI-assisted development. Emphasis on human-in-the-loop validation over speed.
3/11/2026
I Have 30 Years of Career Left. AI Made Me Rethink All of Them
Career reflection on 30-year horizon amid AI disruption. Opinion piece on industry reinvention, not technical.
3/11/2026
AI and the Mixed-Consistency Future
Analysis of file-based storage trends for AI agent memory and RAG systems versus traditional databases. Emerging pattern in AI tooling.
3/11/2026
Show HN: Manage Cursor agents from your smartphone
RebootX tool to manage Cursor agents from smartphone. Extends cloud provider management to AI agents.
3/11/2026
AI productivity gains are 10%, not 10x
Brief claim that AI productivity gains are 10% rather than 10x without supporting analysis or data.
3/11/2026
Decision Guardian: My First GitHub Action and CLI Project
Decision Guardian: GitHub Action and CLI tool that surfaces architectural decisions in PRs to prevent knowledge loss when senior engineers leave.
3/11/2026
Get free ChatGPT Pro for open-source maintainers
OpenAI announces $1M Codex Open Source Fund providing API credits and ChatGPT Pro access to eligible open-source maintainers.
3/11/2026
A practical technique for issue resolution with agentic AI
Analysis/Implementation/Reflection pattern for agentic AI issue resolution: agent explores problem, creates evaluation harness, reflects on solution quality.
3/11/2026
Show HN: HDC-based function caller ranks #2 on BFCL V4 – $2.08 vs. Opus at $87
Glyphh: HDC-based function caller ranks #2 on BFCL V4 leaderboard using hyperdimensional computing for deterministic, sub-millisecond function routing with open-source code.
3/11/2026
SuperPowers: Agentic skills framework that works
Framework for building agentic skills. Skills framework for AI agents, minimal details provided.
3/11/2026
Show HN: An offline-first expense tracker on Cloudflare D1 and SQLite WASM
Expense tracker using SQLite WASM and Cloudflare D1. Offline-first web app, not AI/ML focused.
3/11/2026
The AI Is the Computer
Perplexity Computer product launch explanation discussing AI as primary computing interface without detailed technical specifications.
3/11/2026
AITutor – vimtutor, but for AI-assisted coding
AITutor: Interactive terminal tutorial teaching AI-assisted coding concepts including context windows, tools, MCP, and subagents.
3/11/2026
The Agency: Meticulously crafted AI agent personalities
Announcement of crafted AI agent personality systems without technical implementation details.
3/11/2026
Rabbit r1 with whatever model you want
Framework to run multiple LLM models on Rabbit R1 voice assistant. Supports local offline execution via local-voice-ai.
3/11/2026
Request Copilot code review from GitHub CLI
GitHub CLI integration for Copilot code review. LLM-powered developer tool for terminal-based workflows.
3/11/2026
WordPress/PHP-AI-client: provider agnostic PHP client SDK to communicate with AI
PHP SDK client library providing provider-agnostic interface to communicate with multiple AI models.
3/11/2026
Show HN: CAS – I reverse-engineered Claude Code to build a better orchestrator
CAS: Multi-agent orchestrator with persistent memory and context system. Reverse-engineered Claude Code, manages parallel agents in isolated worktrees.
3/11/2026
Browserbase Fetch: Simple API for extracting web page content for AI agents
Browserbase Fetch API: Simple endpoint for extracting web page content for AI agents without browser session, priced at ~$1 per 1,000 pages.
3/11/2026
Show HN: Email inbox for your OpenClaw agent
MailboxKit: Email API designed specifically for AI agent self-registration and programmatic email operations.
3/11/2026
Meta acquires Moltbook, the AI agent social network
Meta acquires Moltbook, a social network platform for AI agents.
3/11/2026
Music Maker: Automate Music Generation with Claude Code
Tool for automating music generation using Claude Code API.
3/11/2026
Claude Code model comparison: Skill usage
Comparison analysis of Claude Code model skill usage patterns.
3/11/2026
Show HN: Portal (SPC F25) – Try products with browser session sandboxes
Portal: Browser sandbox platform for sharing interactive product demos via shareable links with session replay.
3/11/2026
Microsoft BitNet – inference with 1.58 bits
Microsoft BitNet update: 1.58-bit inference optimizations including parallel kernel implementations and native I2_S GEMM/GEMV support for CPU performance.
3/11/2026
Your AI-generated password isn't random, it just looks that way
Security analysis showing Claude, ChatGPT, and Gemini generate weak passwords despite appearing complex.
3/11/2026
Show HN: Saguaro: CLI that makes Claude Code fix its own mistakes
Saguaro: CLI daemon that reviews AI-generated code and feeds findings back to Claude agents for self-correction.
3/11/2026
Show HN: Nexus – A 1KB .NET 10 core with <15μs latency for 1M parameters
Nexus: Open-source .NET 10 core achieving 12.8μs latency for 1M parameter indexing with zero-allocation architecture.
3/11/2026
Tenets of Agentic Reliability Engineering
Framework defining governance, authorization, and reliability standards for autonomous AI agents deployed in production.
3/11/2026
Show HN: Shadowscan – see what an AI agent can access on your machine
Tool to evaluate what permissions and data an AI agent can access on a user's machine.
3/11/2026
Show HN: Daub – A rendering spec for AI-generated UIs (two files, no build step)
DAUB is a rendering spec for AI-generated UIs using structured JSON with 76 components and zero build step, enabling AI to output UI directly.
3/11/2026
Slicing an 80B MoE LLM into 40B domain specialists
College of Experts framework slices 80B MoE LLM into 40B specialists using Ollama and ONNX-based Supervisor, runs on consumer hardware without CUDA.
3/11/2026
Nvidia Releases NemoClaw – Enterprise AI Agents, Redefined
Ask HN discussion seeking tools to prevent or block AI-generated GitHub issues in public repositories.