AI Developer Tools & SDKs
-
Forge guardrails take an 8B self-hosted model from 53% to 99% on agentic tasks
Forge is an open-source guardrail framework from antoinezambelli that takes an 8B self-hosted model from 53% to 99% accuracy on agentic workflows — within 1 percentage point of frontier APIs running the same framework. It showed up on Show HN this week and was presented as a demo at CAIS 2026. ## What’s in the… Continue reading
-
andrej-karpathy-skills hits 1,955 daily stars: a CLAUDE.md that stops AI from breaking your code
Forrest Chang’s andrej-karpathy-skills is the #1 trending repo on GitHub today with 1,955 stars in a single day — and over 220,000 combined across his personal account and the multica-ai organization mirror. It’s a single CLAUDE.md file encoding four behavioral rules derived from Andrej Karpathy’s documented frustrations with LLM coding agents. ## The core rules… Continue reading
-
Google Antigravity 2.0 ships standalone desktop app, CLI, SDK, and Managed Agents in the Gemini API
Google turned Antigravity into a full agent-first development platform at I/O 2026. Antigravity 2.0 is now a standalone desktop application built entirely around agent orchestration — plus a CLI, an SDK, Managed Agents in the Gemini API, and enterprise support via the Gemini Enterprise Agent Platform. ## The four surfaces Desktop app: a central home… Continue reading
-
Anthropic opens the official Claude Code plugins directory with Anthropic Verified badges for high-trust extensions
Anthropic just opened claude-plugins-official — a curated directory of high-quality plugins for Claude Code and Claude Cowork, managed directly by Anthropic. The marketplace ships built-in to every Claude Code install. ## What’s inside Two top-level directories: `/plugins` for internally-developed Anthropic plugins (code review, SDK helpers, language server integrations, external-service connectors), and `/external_plugins` for third-party submissions… Continue reading
-
pixserp gives LLMs one endpoint, 10 answer shapes, $1.50 per 1k requests
pixserp launched on Product Hunt this week — a single API endpoint that returns 10 different answer shapes (web, news, images, places, shopping, flights, hotels, YouTube, transcripts, any URL) so an LLM can pick the right format for the question instead of stitching together five different services. ## The pricing and architecture $1.50 per 1,000… Continue reading
-
Agentspan ships MIT-licensed durable runtime for AI agents, built on the orchestration engine Netflix uses
Agentspan is an open-source durable runtime for AI agents, MIT licensed, just launched on Product Hunt. The pitch: zero-configuration durable execution for agent workflows. ## What durability means here Step 7 of a 10-step agent workflow crashes? Execution resumes at step 8 — automatically. No checkpointer to set up, no database to configure, no manual… Continue reading
-
humanlayer’s 12-factor-agents resurfaces on GitHub trending: a methodology for production LLM apps
humanlayer’s 12-factor-agents repo climbed back onto GitHub trending today with 359 daily stars. It’s a manifesto plus reference material for building LLM-powered software that’s actually production-grade — modeled on Heroku’s original 12-Factor App methodology. ## The core thesis “Agents, at least the good ones, don’t follow the ‘here’s your prompt, here’s a bag of tools,… Continue reading
-
DreamServer bundles 13 local AI services into one install — LLM, voice, agents, RAG, no cloud
Light-Heart-Labs (with the OpenClaw Collective) released DreamServer — a full local AI stack that bundles LLM inference, chat UI, voice (STT + TTS), AI agents, workflow automation, RAG, web search, deep research, image generation, and privacy tools into a single installer. No cloud, no subscriptions, no API keys. ## What’s in the box 13 services… Continue reading
-
Semble cuts AI agent code-search tokens by 98% with natural language queries instead of grep
MinishLab released Semble — a code search library purpose-built for AI agents. It returns the exact code snippets an agent needs instead of having grep dump full files into the context window. The result: roughly 98% fewer tokens used per code-search step. Show HN hit 212 points this morning. ## The numbers End-to-end indexing of… Continue reading
-
CLI-Anything turns any software into agent-native CLI through a 7-phase Claude Code plugin
HKUDS shipped CLI-Anything — a Claude Code plugin that automatically generates production-ready CLI interfaces for any software with a codebase, turning GUI apps and SaaS platforms into agent-native tools. It’s sitting at 306 daily stars on GitHub trending today. ## The 7-phase pipeline Analyze → Design → Implement → Plan Tests → Write Tests →… Continue reading
Top AI Product 