Developer tools

WorkOS reports security breach (opens news.ycombinator.com)

A WorkOS customer posted a breach notification they say they received, for an incident detected on October 30; it was not yet on WorkOS's website or trust center.

Anthropic clamps down on Claude’s AI video game clones (opens theverge.com)

Anthropic has updated Claude Code's system prompt to prevent the AI tool from being used to re-create existing video games, as reported by Tom's Hardware. Retro gaming is currently being overrun with AI-made decompilations, recompilations, and ports, but while Anthropic will still let people use…

Anthropic’s message threads let Claude agents skip resending chats (opens thenextweb.com)

Anthropic launched message threads in public beta for its API. Agents send the full conversation once, then only new messages, which Anthropic says cuts the wait before Claude replies by about 35% on very long sessions. Prices and rate limits stay the same.

Show HN: OmniDiff – Opensource, super fast, content aware Git difftool, in Rust (opens github.com)

OmniDiff diffs code in 25 languages plus config, images, PDFs, archives and fonts, and is written in Rust.

The AI cost of JavaScript frameworks (opens nolanlawson.com)

agentmash (opens producthunt.com)

open-source multiplayer AI for teams' coding agents

Anthropic takes steps to block Claude from reverse-engineering video games (opens mashable.com)

Game developers report Claude Code refusing work on projects that recreate existing games since late October, some of which they say they have a right to build.

Mirrors: Fix recurring AI agent mistakes with real-time guidance (opens ycombinator.com)

Stop playing whack-a-mole with your agents

Backdoor in Claude Code for VS Code extension on the Cursor Marketplace (opens phorion.io)

A tampered copy of the Claude Code extension was published on Cursor's marketplace. Anthropic says a malicious actor used access to that marketplace and that its own systems and employee laptops were not compromised.

AI isn't replacing SREs, it's deskilling them (opens newsletter.signoz.io)

Google X co-founder Sebastian Thrun raises $130M for AI engineering startup Dulo (opens siliconangle.com)

Dulo, co-founded by Sebastian Thrun, raised $130 million; The Information broke the news and co-founder Adrian Aoun confirmed it.

What Happens When Everyone Can Build Software (opens youtube.com)

Supabase co-founder Paul Copplestone on the Lightcone; the company has 13 million users and over 50 million databases.

Cisco gives engineer unlimited AI tokens to rewrite legacy code (opens fortune.com)

Hue: Simulations for AI agents (opens ycombinator.com)

Hue lets teams test an AI agent in stateful clones of the apps it works in.

Launch HN: Simantic (YC F26) – Firmware Simulation for AI Agents (opens news.ycombinator.com)

Simantic, founded by former firmware engineers from Neuralink and Tesla, builds a simulation layer so engineers and AI agents can test firmware without hardware.

Composal: A swarm of agents that tests your product like real users (opens ycombinator.com)

A swarm of agents that tests a product the way real users would, to catch bugs before users do.

Octoverse: Commits, pushes, reach new highs in 2026 as developers build with agents (opens github.blog)

See how a record year of growth reached every industry, every field, and every kind of builder in the 2026 Octoverse report.

Sources: AI coding startup Cognition weighs a new funding round before year's end after receiving offers valuing it at $70B-$80B, up from $48B in September (opens businessinsider.com)

Sources: AI coding startup Cognition weighs a new funding round before year's end after receiving offers valuing it at $70B-$80B, up from $48B in September — Just weeks after announcing a new round of funding at a $48 billion valuation, AI coding startup Cognition is fielding new investor interest…

Apple's git still trusts the compromised eMudhra certificates (opens emudhra.we-have-a-problem.com)

The Agent SDK (opens cursor.com)

The October inflection point for vibe game development (opens simonwillison.net)

I've written before about how November 2025 was an inflection point for coding agents - the point where Opus 4.5 and GPT-5.1 crossed a tipping point that meant their respective coding agents became reliable daily drivers for software engineers.

Hex Raises $100M (opens hex.tech)

Rystic: Test your AI agents with local copies of Slack, Stripe, and Google Drive 🧪 (opens ycombinator.com)

Test your agent in a simulated world before unleashing it on the real one.

Former GitHub CEO launches rival platform built for AI coding agents (opens geekwire.com)

Entire's Trails feature replaces the pull request, using AI agents to score each code change.

I checked the DNS of 8,062 Show HN launches. 1 in 5 send email with Resend (opens reddit.com)

I took every product posted to Show HN in the 120 days before Oct 13 (14,880 posts, 8,062 unique domains) and checked their public DNS. - 1,580 of them send email with Resend, about 1 in 5. One DNS lookup tells you, since Resend signs with resend._domainkey.

Why Reddit built its own AI prototyping tool | Blair Metcalf (opens lennysnewsletter.com)

Watch now (30 mins) | 🎙️ Reddit’s principal UX engineer shows how a custom Figma MCP and Cursor pipeline takes any designer from blank slate to shareable, production-shaped prototype in minutes—no Git required

Pomerado - The faster, cheaper, and more reliable alternative to browser use (opens ycombinator.com)

YC startup Pomerado connects agents to any website without needing an API, and it is open source.

Show HN: Farol – A risk map of a pull request for the person who approves it (opens github.com)

Same game built twice: Opus alone vs Opus planning + local Qwen coding. 16 min vs 42 min, $2.52 vs $0.62 (opens reddit.com)

The poster kept Opus 5.5 for planning and moved code writing to a local Qwen model; their reported result was 42 minutes and $0.62 versus 16 minutes and $2.52 with Opus alone.

Show HN: ADRs for AI – Deliberation records for human–AI work, versioned in Git (opens adrs.systems)

Why is this worth your time? Essentially, this solves two problems of AI-assisted programming: (a) the first problem is how to do oversight and what do oversight of and this is solved through the ORRCF (for Options, Recommendation, Rationale, Confidence, Falsifier) question format (in which models…

Roku developer repositories moved from GitHub to GitLab (opens blog.roku.com)

Source: Replit recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year (opens theinformation.com)

Source: Replit recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year, after Anthropic discontinued its discount — OpenAI's efforts to undercut Anthropic on price are producing some results. — For instance, in May, Anthropic models powered 57% …

Anthropic’s new guide (Oct 22) argues most agents should have fewer subagents: one main loop, skills and tool search instead. Includes “we cut Claude Code’s system prompt by over 80%” (opens reddit.com)

Anthropic published an 18-minute guide on Oct 22 called "When to fold subagents into your main agent loop." It is aimed at enterprise agent builders, but the lessons apply to anyone who set up a pile of subagents in Claude Code.

Hindsight puts up to $1M behind its AI content moderation as Google shuts down Perspective API (opens thenextweb.com)

Google's Jigsaw is shutting down the Perspective toxicity-scoring API on December 31, and Hindsight is offering up to $1M to move users to its AI moderation.

Kimi K3.1 Pro and K3.1 Pro Highspeed (opens kimi.ai)

Moonshot AI's upgraded open-weight model; The Next Web reports it nearly matches Claude Opus 5.5 on a coding test, with full weights due by 4 November and hacking-related training data removed.

Show HN: Corgi, a helpful {agent, terminal} multiplexer (opens corgi.computer)

Hey everyone, Inspired by Herdr, I've built an agent multiplexer that I've been using as my daily driver for a few weeks now. I think I improved on the concept in a few ways, focusing on the ergonomics I always wished I had: I show you what the agents are up to, along with a running feed of their…

LLMs are making formal software verification feasible at a much larger scale (opens anthropic.com)

Agent Relay - Let your teams' coding agents message each other (opens ycombinator.com)

Real time messaging for Claude Code, Codex, Cursor, Conductor or any other agent on your team.

Build more responsive apps with the Decisions API (opens openai.com)

Fast, focused, low-cost decisions for any application.

Simulithic — AI users that test your app on every device and browser. (opens ycombinator.com)

Simulithic (YC Fall 2026) runs AI users against an app on every device and browser.