AI Developer Tools & SDKs
-
Agentic chip design goes autonomous at DAC 2026 (Synopsys / Cadence / Siemens)
At DAC 2026 the three companies that own chip design software all said the same thing on the same week: the AI stops assisting and starts driving. These are software agents that design and verify chips — not the chips themselves. What each one shipped Synopsys put out a fully autonomous verification agent, built on… Continue reading
-
openwork clones Claude Cowork as a self-hosted agent — +97 GitHub stars in a day
Anthropic shipped Cowork, an agent that grinds through your email, calendar, and files in the background — close the laptop and it keeps going, no device online. The community’s answer arrived fast. openwork, from the different-ai team, is an open-source, self-hostable take on the same idea, and it’s picking up +97 stars a day as… Continue reading
-
Hugging Face speech-to-speech (open-source local voice agents) is the OpenAI Realtime clone you can run on your own GPU
OpenAI’s Realtime API gives you a voice agent, but you rent it and your audio leaves the building. Hugging Face just shipped the open-source answer: speech-to-speech, a framework that chains VAD (Silero v5) → STT → LLM → TTS into one parallel pipeline and exposes a WebSocket API that’s drop-in compatible with OpenAI Realtime. Point… Continue reading
-
book-to-skill turns your PDF textbooks into Claude Code skills — 12.6k GitHub stars and counting
Your technical books are dying as forgotten PDFs. book-to-skill is an open-source CLI that fixes that: feed it a PDF or EPUB, and it spits out a structured knowledge base your coding agent can actually query. It gained 1,428 stars in a single day, 12.6k total, because it hits a raw nerve in the coding-agent… Continue reading
-
TurboFieldfare — Gemma 4 26B running in ~2GB RAM on any M-series Mac
A 26B model in 2GB of RAM sounds impossible. That’s exactly why it hit 556 points on Hacker News and 865 GitHub stars in a day. What it actually is TurboFieldfare is a local inference engine written in Swift and Metal that runs Google’s Gemma 4 26B-A4B — 14.3GB of weights — on an 8GB… Continue reading
-
Fireworks Nexus routes routine coding to open-weight models, cutting AI bills 50–75%
Fireworks AI shipped Nexus on July 28, a drop-in routing and cost-control layer that sits under your existing coding agents and decides, per request, whether a task actually needs a frontier model. Most of them don’t. Routine edits go to an open-weight model; the hard stuff still passes through to Claude or GPT on your… Continue reading
-
OpenAI open-sources Codex Security (CLI + TypeScript SDK), hits 138 points on HN in a day
The security agent OpenAI used to keep behind ChatGPT Pro and Enterprise is now Apache-2.0. Codex Security is a command-line tool plus a TypeScript SDK that reads your repo, finds vulnerabilities, checks whether they’re real, and writes the patch. Point it at code, get back fixes. What it actually does It’s an AI coding agent… Continue reading
-
Model Context Protocol 2026-07-28 spec (final): MCP drops sessions, tools get a UI
MCP is the open standard Anthropic built so AI agents can talk to tools and data. The 2026-07-28 spec just went final, and it’s the biggest rewrite since launch. What actually changed The core went stateless. No more session affinity, which means you can park an MCP server behind a dumb round-robin load balancer and… Continue reading
-
Moltbook joins Meta Superintelligence Labs, and OpenClaw ships v2026.7.1
A social network where the users are all bots just got bought by Meta. Moltbook — the “Reddit for AI agents,” where only OpenClaw bots post, comment, and vote while humans can just watch — is now part of Meta Superintelligence Labs. Founders Matt Schlicht and Ben Parr are joining the team. What these two… Continue reading
-
Kimi AgentENV: Moonshot open-sourced the agent-RL training layer nobody else shares
On K3 Open Day, everyone stared at Kimi K3 — the 2.8T-param open model. The quieter drop matters more if you build infra: Kimi AgentENV, the distributed system that actually trains agent models, now open-sourced with kvcache-ai. What it actually is AgentENV isn’t a model. It’s the ground floor under one — a platform for… Continue reading
