Short-form updates on what we're building, learning, and shipping at Kraus Labs.
martinfowler Sensible Defaults + Fragments; cursor_ai /visualize charts; poteto Grok Bot in ChatGPT; swyx podcast ranking + pavilion; hwchase17 org harness; rauchg AI SDK 30M; simonw DevDay liveblog; HamelHusain Anthropic evals; GH: VoiceStudio, OpenShell, hindsight, paperclip, dbx.
X: cursor_ai Opus 5.5 in Cursor; poteto Grok Bot perf; rauchg Next.js evals + software forever; simonw/swyx/Hamel on Opus 5.5 & evals; GH: anthropics/financial-services, google/ax, davila7/claude-code-templates, BuilderIO/agent-native, obra/superpowers.
poteto on statecharts plus formal verification, rauchg on durable-harness, HamelHusain on decision models for evals, swyx on DeepSeek serving margins; MadsLorentzen/ai-job-search tops GitHub trending.
Gemini 4 Argon in Cursor, poteto on banishing code comments and bot-to-bot email, hwchase17 on harness routers; morluto/rea tops GitHub trending.
Cursor agents from the iOS app; poteto automates release and QA with Grok Bot Team Bots; Vercel AI Gateway confidence fallback; simonw on Mistral Large 4 and LLM music; GH: e2e, mattpocock/skills, text-to-cad, AnyPS5, impeccable.
cursor_ai Sonnet 5.5; poteto restate prompt + team bots + finance; swyx OpenAI teaser; hwchase17 Jev retrieval + harness; rauchg F1 migrate + domain search; simonw Sonnet free tier + keynote; HamelHusain Anthropic evals; GH: VoiceStudio, paperclip, hindsight, PLFM_RADAR, coursebook.
X: poteto (2.5k PRs talk, Grok 4.7), rauchg (Grok 4.7/Vercel + Jev HTTP), hwchase17/simonw/Hamel on decision models/Jev; GH: anthropics/financial-services, agent-substrate/substrate, dream-num/univer, davila7/claude-code-templates, google/ax.
Cursor ships its Agent SDK and adds DeepSeek V4.1 Flash, rauchg reproduces covert CPU communication between microVMs, HamelHusain defends benchmark reviews; Nutlope/hallmark tops GitHub trending.
poteto asks what people shipped with Grok Bot, rauchg open sources a durable libfx HTML artifact builder, hwchase17 on vertical agents; MadsLorentzen/ai-job-search tops GitHub trending.
Cursor joins The X Subscription, poteto on the codebase as agent memory, hwchase17 on async subagents, rauchg on Meta Muse Wonder and fx; morluto/rea tops GitHub trending.
Cursor SDK run.steer(); poteto /poteto-help + Grok Bot 101; hwchase17 on agent memory cleanup; rauchg gdp-ts; GH: e2e, claude-mem, text-to-cad, t3code, AnyPS5.
Cursor Rollouts + 7% token cut; poteto Grok Bot studio; rauchg agent triad; simonw Gemini TTS; Hamel evals; GH: financial-services, google/ax, claude-code-templates, agent-native, superpowers.
rauchg on AI explainers and Vercel Node.js RCE mitigations, poteto on Grok Bot Pages, hwchase17 on fine-tuning from traces, HamelHusain on BenchmarkGate; cloudflare/security-audit-skill tops GitHub trending.
poteto on pstack principles and Grok Voice in Grok Bot, rauchg on agent containment and Vercel Ship; morluto/rea tops GitHub trending.
Claude Haiku 5.5 lands in Cursor; poteto on Grok Bot setup and a time-to-rewrite heuristic; rauchg on knowing when agents should stop; Deep Agents skill-scoped tools; GH: rea, mattpocock/skills, AnyPS5, i-have-adhd, diagram-design.
martinfowler slides principles; cursor_ai Innate robots; poteto Grok Bot+cloud agents + burnout; swyx media attribution + Flow; rauchg Connect + AI Gateway; HamelHusain Auto Eval review; GH: OpenShell, VoiceStudio, openrig, context-mode, ponytail.
poteto Opus typo + summon meme; rauchg slop grenades + Lisa Rauch birth; simonw Kākāpō pixel-art video; HamelHusain synthetic data + record problems; GH: paperclip, hindsight, Model-Optimizer, univer, tensorflow.
Today's notable activity from X, GitHub trending repos, and the builder community.
rauchg on over- vs under-guiding AI, poteto on pstack and bots as teammates, hwchase17 on the five components of a company harness; cloudflare/security-audit-skill leads GitHub stars again.
poteto works on a new pstack verification skill, swyx offers TypeSafe referrals, rauchg quotes Tim Sweeney on the end product; anthropics/claude-plugins-official tops GitHub trending.
poteto on codebase slop, 16 parallel Projects, smaller pstack; rauchg sci-fi in our lifetime; HamelHusain simulations for evals; GH: rea, AnyPS5, artcraft, diagram-design, context-mode.
poteto Coding is dead article + chalant/Starship; hwchase17 Jev+LangGraph; rauchg skills.sh + Klaviyo + Jev Pokemon; simonw industry pace + keynote; HamelHusain eval UX + tier list; GH: paperclip, claude-plugins-official, hindsight, superpowers, mattpocock/skills.
We're building research-grade tooling for ambitious teams — and this is where we'll share what we're learning along the way.
Cursor agents can now control your computer, poteto on Grok Bot 0.75.5 and saved logins, simonw on Kimi K3.1, hwchase17 on skills as memory, rauchg on cross-VM timing attacks; cloudflare/security-audit-skill leads GitHub stars today.
Cursor cloud agents build canvases and get an Environments API, poteto on trusting agent work, swyx launches High Taste Testers, rauchg on durable libfx harnesses; anthropics/claude-plugins-official tops GitHub trending.
poteto agent-era interviews; hwchase17 64% cheaper Open SWE routing; rauchg 60%+ agentic Vercel deploys; simonw voice coding; GH: rea, AnyPS5, skills, diagram-design, open-code-review.
cursor_ai Rollouts auto-fix; poteto product layout + 2,500 PRs + Grok Bot; swyx dots + AI Security; hwchase17 verifier context + Custom Apps + memory; rauchg Jev Python + Muse Gadgets; HamelHusain remote agents + evals interview + binary labels; GH: Agent-Reach, caveman, superpowers, ponytail, impeccable.
poteto Grok Bot podcast; rauchg Mini browser Rust/Swift port; HamelHusain Quail + Jev evals + human agents; GH: paperclip, hindsight, VoiceStudio, ai-engineering-from-scratch, PipePipe.
X: poteto, rauchg, hwchase17, swyx, Hamel on Jev/agents; GitHub: agent-native, cua, OpenStock, ai-memory, coder.
poteto on a yminsky talk about making codebases agent-friendly and on languages for agents, HamelHusain on public benchmark quality; makecindy/cindy tops GitHub trending.
poteto ships a pstack formal verification skill, swyx says data engineers are in higher demand than AI engineers, rauchg highlights AI Gateway agent sessions; MadsLorentzen/ai-job-search tops GitHub trending.
Karpathy on the widening LLM capability funnel, poteto on clean code and agent verification loops, hwchase17 on agent architecture and identity; morluto/rea tops GitHub trending.
Fowler Fragments; poteto Ship Company + managing agents; hwchase17 agent costs; rauchg Rust at Vercel + human vs agent docs + fx speed; GH: e2e, impeccable, marketingskills, ponytail, text-to-cad.
martinfowler fragments + feedback; poteto Cursor Projects/pstack + Bend; rauchg model spend mix; simonw agents harder; swyx Scaling without Slop; GH: ai-engineering-from-scratch, hindsight, univer, google/ax, Model-Optimizer.
poteto on a statechart companion skill and Grok Bot phone apps, rauchg on Vercel backends and retiring static credentials, swyx on Opus 5.5 work estimates, hwchase17 naming agent servers; cloudflare/security-audit-skill leads GitHub stars today.
Muse Wonder lands in Cursor, poteto on pstack, hwchase17 on harness meets infra, rauchg on Vercel backends, Drives and Daemons; morluto/rea tops GitHub trending.
poteto on PR volume with agents; hwchase17 on evals vs Goodhart and Jev trajectory labeling; simonw on Microsoft cross-platform agent sandbox; GH: AnyPS5, diagram-design, rea, skills, claude-mem.
cursor_ai GLM 5.3; karpathy Land/Water eval + LLM output tips; poteto Grok Bot; hwchase17 durable runtime + model routing + decision models; rauchg verification-engineering + SvelteKit 3 + MicrosoftAI; HamelHusain eval costs; GH: ponytail, skills, OpenShell, firebase-ios-sdk, openrig.