Agents Hour

Dex Horthy has spent the past year studying what happens to a codebase when you let AI write all of it and stop reading the output. His answer has a name: slop. And it's not just about sloppy prompting.
In this episode, the HumanLayer founder Dex Horthy joins Shane and Abhi to draw the software factory on a whiteboard — the whole pipeline that builds the thing that builds the thing — and show exactly where teams tear out the parts that were holding quality up. He moves from the naive factory to the fully agentic one to the "lights-off" version where humans stop reading code entirely, then explains why that dream quietly rots a codebase after three to six months.
Dex, Shane, and Abhi also talk about how coding models are trained. Every benchmark rewards one thing: did the test pass? Nowhere in the loop does a model get punished for leaving behind code that a smaller model, or a human six months later, can't work in. That gap is why architecture planning, sprint planning, and code review still can't be skipped.
You'll also hear why you're not drowning in PRs so much as drowning in bad ones, what a longer-horizon benchmark that actually catches slop would look like, and how many times Dex has watched the HumanLayer launch film.
🔗 DEX HORTHY / HUMANLAYERDex on X: https://x.com/dexhorthyHumanLayer: https://humanlayer.dev
AI Agents Hour is a weekly livestream by Mastra CPO Shane Thomas and CTO Abhi Aiyer. Mondays 12PM Pacific.
📚 MASTRA RESOURCESMastra: https://mastra.aiMastra on X: https://x.com/mastra_aiMastra Discord: https://mastra.ai/community/discordMastra GitHub: https://github.com/mastra-aiLearn Mastra: https://mastra.ai/learnPrinciples of Building AI Agents: https://mastra.ai/books/principles-of-building-ai-agentsPatterns for Building AI Agents: https://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRAMastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
CHAPTERS0:00 Cold open0:14 Bringing on Dex Horthy1:16 The HumanLayer launch film3:31 Cutting through the AI hype4:07 Defending the terms that matter5:26 "I want to build a software factory"5:58 What a software factory actually is9:08 Agentic code review, testing, and feedback loops13:35 The lights-off software factory14:32 Why code quality decays over time15:36 What RL actually rewards17:29 Don't sacrifice planning or review18:05 Getting the PR to 95% before review19:21 Drowning in bad PRs, not PRs20:22 Building longer-horizon benchmarks21:46 Trying HumanLayer

GPT-5.6 is here — Sol, Terra, and Luna — but for the first time, the US government decides who gets access, approving customers one by one. Shane and Abhi dig into what that precedent means for competition, open models, and every lab that comes next.
Then a stacked week: OpenAI's own Jalapeño chip, open models pushing hard (Qwen AgentWorld, DeepSeek DSpark, Hermes mixture-of-agents), the model-routing wave (AI SDK 7, Devin Fusion), Claude Tag's viral moment and the Trojan-horse worry, voice agents on Vercel, Coinbase on keeping AI spend flat, a16z's $200M "seed" for Mirendil, npm finally freezing accounts after the Axios and Mastra attacks, Meta's think-to-text Brain2Qwerty, the DevRel Google fired over a viral CLI, and Fable's return after 15 days offline.
AI Agents Hour is a weekly livestream by Mastra CPO Shane Thomas and CTO Abhi Aiyer. Mondays 12PM Pacific.
📚 READ MOREGPT-5.6 Sol/Terra/Luna: https://x.com/OpenAI/status/2070555272230384038Restricted rollout: https://x.com/steph_palazzolo/status/2070241787180966279Jalapeño chip: https://x.com/OpenAI/status/2069770172802773292Qwen AgentWorld: https://x.com/Alibaba_Qwen/status/2069720365442719867DeepSeek DSpark: https://x.com/Yuchenj_UW/status/2070928299744972814Hermes Mixture of Agents: https://x.com/NousResearchAI SDK 7: https://x.com/vercel/status/2070155382488764566Devin Fusion: https://x.com/cognition/status/2071624568465490170Tau harness: https://x.com/_alejandroaoClaude Tag (Karpathy): https://x.com/karpathyClaude Tag Trojan horse: https://x.com/ashwingopOpen Tag: https://x.com/ataiiamGLM-5.2 in the wild: https://x.com/cline/status/2069171146994729078Local 1-bit GLM: https://x.com/UnslothAI/status/2069418532375564484Coinbase on AI spend: https://x.com/brian_armstrongCoinbase gateway: https://x.com/markletreeExa Connect: https://x.com/ExaAILabs/status/2069842203577651283Mirendil $200M: https://x.com/a16z/status/2069869327411749012npm account freeze: https://x.com/AikidoSecurityBrain2Qwerty: https://x.com/AIatMeta/status/2071566924803395741Google Workspace CLI: https://x.com/JPoehnelt
📚 MASTRA RESOURCESMastra: https://mastra.aiMastra on X: https://x.com/mastra_aiMastra Discord: https://mastra.ai/community/discordMastra GitHub: https://github.com/mastra-aiLearn Mastra in the world's first MCP-Based Course: https://mastra.ai/coursePrinciples of Building AI Agents (Book): https://mastra.ai/books/principles-of-building-ai-agentsPatterns for Building AI Agents (New Book): https://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRA?Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development—from prototype to production. You can integrate it with frontend and backend stacks (e.g., React, Next.js, Node) or run agents as standalone services. If you're a JavaScript or TypeScript developer looking to build an agentic or AI-powered product without starting from first principles, Mastra provides the scaffolding, tools, and integrations to accelerate that process.
⏱️ CHAPTERS00:00 Cold open: npm finally freezes accounts00:29 Welcome00:37 GPT-5.6 Sol, Terra, Luna — and government-gated access04:25 OpenAI's Jalapeño chip05:34 Open models: Qwen AgentWorld, DeepSeek, Hermes MoA07:11 AI SDK 707:43 Devin Fusion & the model-routing wave10:26 Harness interop: OpenCode & LangChain11:01 Tau harness11:26 Claude Tag's viral moment & the Trojan horse12:49 Open Tag13:12 Voice agents land on Vercel13:53 The infra bill: Coinbase on keeping AI spend flat15:15 Exa Connect & a16z's $200M "seed"16:32 npm freezes accounts18:08 Meta's Brain2Qwerty: think-to-text19:11 Google fired its viral CLI star21:48 Fable returns22:00 Outro

Mastra got hacked.
In this special edition of Security Corner, Shane Thomas and Abhi Aiyer break down exactly what happened when a supply chain attack hit Mastra's npm packages — an attack that appears to trace back to hackers in North Korea.
They're joined by Ismail Pelaseyed, co-founder and CTO of Superagent, for the outside view on how these campaigns actually work.
You'll hear how a single, ordinary-looking call turned into a full npm account takeover, the small oversight that turned a scare into a genuine crisis, and why a malicious package was still live on the registry weeks after it was reported.
Ismail makes the case that getting hacked is a side effect of success — and that the real problem runs deeper than any one team. You'll learn why he thinks npm and PyPI have dropped the ball on security, how AI now lets attackers one-shot a convincing phishing app, and what every maintainer should be doing to harden their pipeline before a trusted contributor becomes the way in.
It's the unfiltered version, told by the people who lived through it.
Connect with Ismail Pelaseyed:https://x.com/pelaseyedhttps://superagent.sh
Connect with the hosts:https://x.com/smthomas3https://x.com/abhiaiyer
📚 MASTRA RESOURCEShttps://mastra.aihttps://x.com/mastra_aihttps://mastra.ai/community/discordhttps://github.com/mastra-aihttps://mastra.ai/coursehttps://mastra.ai/books/principles-of-building-ai-agentshttps://mastra.ai/books/patterns-of-building-ai-agents
Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
CHAPTERS0:00 Intro: a special Security Corner0:55 The supply chain attack on Mastra1:29 How they got in: a fake Teams call2:27 The npm account takeover3:54 EasyDjS and the scramble to fix it5:49 Why success makes you a target9:20 How AI supercharges phishing9:59 Hardening against compromised contributors11:10 Open source under strain: IBM's $5B bet12:27 npm and PyPI keep dropping the ball14:31 Inside the fake package, and how Socket caught it16:20 The fear-selling problem in security18:02 Superagent!

Fable is gone, and the race to replace it is already on. Shane and Abhi open with the fallout — reports that Mythos breached classified NSA systems, the ID-verification future that may follow, and David Sacks' "you asked for it" take.
Then the story flips: OpenAI's GPT-5.5 Cyber lands state-of-the-art on Cyber Gym, beating Fable and Mythos on the underlying benchmarks, with OpenAI Daybreak shipping alongside it. GPT-5.6 leaks, then slips to mid-July. GLM 5.2 arrives near Opus 4.8, tops Design Arena, and runs on a Mac. Cursor goes after GitHub with Origin, adds mobile and its own frontier model on Colossus after the SpaceX deal closes. Vercel ships Eve, Fred Schott counters with Flue, Claude lands in Slack, and Midjourney — yes, the image company — unveils a full-body medical scanner headed for spas.
AI Agents Hour is a weekly livestream by Mastra CPO Shane Thomas and CTO Abhi Aiyer. Mondays 12PM Pacific.
📚 READ MOREThe Fable story, Mythos & the NSA: https://x.com/kimmonismus/status/2068605229965234238GPT-5.5 Cyber beats Fable (Sam Altman): https://x.com/sama/status/2069121360744550796OpenAI Daybreak: https://x.com/OpenAI/status/2069104283824640023GPT-5.6 leak (Mark K): https://x.com/mark_k/status/2069169916432003105GPT-5.6 delayed: https://x.com/synthwavedd/status/2069432791184650426GLM 5.2 (Z.ai): https://x.com/zai_org/status/2066938937344495629GLM 5.2 tops Design Arena: https://x.com/Designarena/status/2066940737011560652GLM 5.2 vs Opus (Cline): https://x.com/cline/status/20691711469947290781-bit GLM 5.2 runs local (Unsloth): https://x.com/UnslothAI/status/2069418532375564484Cursor Origin: https://x.com/cursor_ai/status/2069149296436330776Cursor Compile keynote: https://x.com/cursor_ai/status/2067012220832329782Swyx on Origin: https://x.com/swyx/status/2066928345246470204Vercel Eve: https://x.com/vercel/status/2067180054979936413Flue 1.0 (Fred Schott): https://x.com/FredKSchott/status/2066962296119959581Codex record-replay to skills: https://x.com/OpenAIDevs/status/2067681320281723113Claude Tag in Slack: https://x.com/claudeai/status/2069468693017268244Replit in Slack: https://x.com/Replit/status/2067661213278900402Midjourney Medical: https://x.com/midjourney/status/2067421950314688759Sakana Fugu: https://x.com/SakanaAILabs/status/2068861630327443966
📚 MASTRA RESOURCESMastra: https://mastra.aiMastra on X: https://x.com/mastra_aiMastra Discord: https://mastra.ai/community/discordMastra GitHub: https://github.com/mastra-aiPrinciples of Building AI Agents (Book): https://mastra.ai/books/principles-of-building-ai-agentsPatterns for Building AI Agents (New Book): https://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRA?Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development—from prototype to production. You can integrate it with frontend and backend stacks (e.g., React, Next.js, Node) or run agents as standalone services. If you're a JavaScript or TypeScript developer looking to build an agentic or AI-powered product without starting from first principles, Mastra provides the scaffolding, tools, and integrations to accelerate that process.
⏱️ CHAPTERS00:00 Cold open00:30 Welcome01:02 Fable fallout: did Mythos breach the NSA?03:10 GPT-5.5 Cyber beats Fable & OpenAI Daybreak05:23 GPT-5.6 delayed, Gemini 3.5 Pro slips05:52 GLM 5.2: near-Opus, and it runs on a Mac09:52 Subscribe break09:55 Cursor's Origin takes on GitHub11:12 Cursor mobile, frontier model & the SpaceX deal11:52 Vercel's Eve, Flue & Codex skills13:48 Claude Tag & Replit in Slack14:57 Midjourney goes medical: full-body scanners16:42 Sakana Fugu & the rise of model routing19:37 Outro

Open source is under attack, and AI changed the math. In this Security Corner, Ismail Pelaseyed, co-founder and CTO of Superagent, joins Shane and Abhi to break down how the software supply chain became the soft underbelly of everything we build.
An attack that once took an army of researchers and weeks of work now takes about an hour, and the attacker no longer needs a frontier model to pull it off. Ismail traces how most breaches begin, why phishing has become almost impossible to spot, and how a single poisoned dependency can cascade across an entire ecosystem.
You'll get concrete steps any maintainer or developer can take today: switching package managers, enabling the security scanners that ship for free, and standing up an adversarial agent that hunts for chained exploits before an attacker finds them.
Ismail also warns that the same instincts protecting enterprises may be quietly strangling open source itself. You'll hear why he thinks the big registries have dropped the ball, what a "Darwinian GitHub" would mean for anyone shipping a new package, and the one move he believes can keep the ecosystem alive.
Superagent: https://superagent.shSuperagent on X: https://x.com/superagent_aiSuperagent on GitHub: https://github.com/superagent-aipnpm: https://pnpm.ioSocket: https://socket.dev
📚 MASTRA RESOURCESMastra: https://mastra.aiMastra on X: https://x.com/mastra_aiDiscord: https://mastra.ai/community/discordGitHub: https://github.com/mastra-aiFree course: https://mastra.ai/coursePrinciples of Building AI Agents: https://mastra.ai/books/principles-of-building-ai-agentsPatterns of Building AI Agents: https://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRAMastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
⏱️ CHAPTERS0:00 Cold open0:21 What is Superagent0:53 How AI sped up attack timelines2:14 Why phishing is the way in4:35 Outdated CI/CD workflows6:04 Two defenses: CI/CD checks and switching to pnpm7:18 The risk hiding in skills and agents8:11 Should you delay installing new packages?8:54 The Darwinian GitHub threat to open source9:55 Why supply chain attacks are so popular11:34 Will companies abandon open source?13:53 Why Ismail is frustrated with GitHub and npm14:32 Practical defenses for maintainers18:12 Where to find Superagent

Fable came and went in a week. Shane and Abhi break down the strangest model launch yet — Claude Fable 5, a Mythos-class model Anthropic said was too capable to release widely, until a US export-control directive made it vanish for everyone.
We cover the whole arc: the launch hype, the pricing that had people spending $1,000 a day, the silent fallbacks to Opus, the system-prompt leak, and the snitch that triggered a 90-minute shutdown. Then a stacked back half — China's open-weight surge, OpenRouter Fusion, loop engineering vs loopcraft, the agent-learning gold rush, OpenAI filing to go public, and Dario's new essay.
AI Agents Hour is a weekly livestream by Mastra CPO Shane Thomas and CTO Abhi Aiyer. Mondays 12PM Pacific.
📚 READ MOREClaude Fable 5 launch: https://x.com/claudeai/status/2064394146916229443Pokémon beaten with vision: https://x.com/Baconbrix/status/2064418858073784530Export-control announcement: https://x.com/anthropicai/status/2065597531644743999Amazon/WSJ report (Theo): https://x.com/theo/status/2065665304882209132David Sacks on the ban: https://x.com/davidsacks/status/2065853007619588171Token capital (Satya Nadella): https://x.com/satyanadella/status/2066182223213293753Siri as a harness (Weinbach): https://x.com/mweinbach/status/2065630219021492474MiniMax M3 open weights: https://x.com/minimax_ai/status/2065436935188058208GLM 5.2: https://x.com/zai_org/status/2065704919299235870Kimi K2 Code High Speed: https://x.com/kimi_moonshot/status/2066467110960959833Cohere North Mini Code: https://x.com/cohere/status/2064378058329526556Gemma 4 on consumer hardware: https://x.com/unslothai/status/2065433326706684135Diffusion Gemma: https://x.com/unslothai/status/2065107734916432189OpenRouter Fusion API: https://x.com/openrouter/status/2065856853989270011AI SDK harnesses (Vercel): https://x.com/vercel_dev/status/2065509970775519569Swyx on loopcraft: https://x.com/swyx/status/2065307558198567206OpenEnv joins Hugging Face: https://x.com/ben_burtenshaw/status/2063991191415267492Adaline 2.0 / agent learning: https://x.com/adiix_official/status/2066172819952566643Cognition Frontier Code: https://x.com/cognition/status/2064061031912288715Ramp SWE-bench: https://x.com/RampLabs/status/2065485806605619304AWS on AI-generated code: https://x.com/awscloud/status/2064449711155589396OpenAI files to go public: https://x.com/OpenAINewsroom/status/2065088002335158753OpenAI acquires Ona: https://x.com/OpenAINewsroom/status/2064094175541461220Salesforce acquires Fin AI: https://x.com/fin_ai/status/2066493998852686074Mistral in talks to raise: https://x.com/business/status/2065420995393941692Dario's essay on the AI exponential: https://x.com/DarioAmodei/status/2064781775247950326
📚 MASTRA RESOURCESMastra: https://mastra.aiMastra on X: https://x.com/mastra_aiMastra Discord: https://mastra.ai/community/discordMastra GitHub: https://github.com/mastra-aiLearn Mastra in the world's first MCP-Based Course: https://mastra.ai/coursePrinciples of Building AI Agents (Book): https://mastra.ai/books/principles-of-building-ai-agentsPatterns for Building AI Agents (New Book): https://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRA?Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development—from prototype to production. You can integrate it with frontend and backend stacks (e.g., React, Next.js, Node) or run agents as standalone services. If you're a JavaScript or TypeScript developer looking to build an agentic or AI-powered product without starting from first principles, Mastra provides the scaffolding, tools, and integrations to accelerate that process.
⏱️ CHAPTERS00:00 Cold open & welcome00:40 Claude Fable 5: the launch and the hype01:53 The price problem and silent Opus fallbacks06:42 Our experience + the system-prompt leak08:51 The 90-minute export-control shutdown12:40 Microsoft's "token capital" & where Fable goes17:07 Apple at WWDC: Siri as a harness17:57 China's open models: MiniMax M3, GLM 5.2, Kimi K221:11 Local-first: Cohere, AMD, Diffusion Gemma24:58 OpenRouter Fusion & AI SDK harnesses27:07 Loop engineering, loopcraft & OpenEnv29:09 Agent learning & new coding benchmarks32:14 AWS, the snitch & OpenAI's IPO33:37 Salesforce, Mistral & Le Chaton Fat36:42 Quick hits & a Fable eulogy

Shane and Abhi are in person this week — live from the CodeRabbit office — for a packed AI news rundown.
The big theme: loop engineering. Boris Cherny (head of Claude Code) and Peter Steinberger both landed the same take within days — stop prompting your agents, start designing the loops that prompt them. We walk the whole evolution: the traditional loop, the Ralph loop, /goal, and Claude Code's new dynamic workflows — and debate whether "stop prompting" is real insight or just clickbait.
Plus: Anthropic engineers shipping 8x more code (and the "depressed employees" reply), agentic traffic passing human traffic on the web for the first time, OpenAI's Codex Sites taking aim at Lovable, Cognition's $10M AI Productivity Guarantee and Devin Desktop, Cloudflare acquiring VoidZero, the accelerating shift to Chinese models (Lindy going 100% DeepSeek), Notion disabling Anthropic models over reliability, a big open-model dump (Gemma 4, Magenta RealTime 2, Miso One, Nemotron, Liquid, GLM 5.1, MiniMax M3), funding rounds (Suno, Supabase), Brian Chesky's new AI lab, and whether AI is actually profitable yet.
Recorded live at the CodeRabbit office — thanks to the CodeRabbit team.
AI Agents Hour is a weekly livestream by Mastra CPO Shane Thomas and CTO Abhi Aiyer. Mondays 12PM Pacific.
📚 READ MOREAnthropic ships 8x more code: https://x.com/AnthropicAI/status/2062568864240836995"Depressed employees" reply: https://x.com/jasonbotterill/status/2062579899412713605Bots pass humans (Matthew Prince): https://x.com/eastdakota/status/2062212701414187452Boris Cherny on loops (via @rohanpaul_ai): https://x.com/rohanpaul_ai/status/2063289804708835412Steipete — design loops, don't prompt: https://x.com/steipete/status/2063697162748260627OpenAI Codex Sites: https://x.com/openai/status/2061845949170045346Cognition AI Productivity Guarantee: https://x.com/cognition/status/2062597242167628019Devin Desktop: https://x.com/cognition/status/2061889596703551926Cloudflare acquires VoidZero: https://x.com/voidzerodev/status/2062520542121304146Shift to Chinese models (Nick Thompson): https://x.com/nxthompson/status/2063712713654628549Lindy → DeepSeek V4 (Flo Crivello): https://x.com/altimor/status/2062389885437366342Notion disables Anthropic models: https://x.com/notionstatus/status/2063477745796161904Gemma 4 12B: https://x.com/Google/status/2062203526588088452Gemma 4 QAT (Unsloth): https://x.com/UnslothAI/status/2062931482746994755Magenta RealTime 2: https://x.com/googlegemma/status/2062619217967628693Miso One: https://x.com/aodenteomt/status/2062204362102100295Nemotron-3.5-ASR-Streaming: https://x.com/piotrzelasko/status/2062538923776290909Liquid LFM2.5-VL-Extract: https://x.com/liquidai/status/2062686748291846307Baseten — GLM 5.1 at 160+ TPS: https://x.com/baseten/status/2062942929883426860MiniMax M3 faster: https://x.com/ryanleeminimax/status/2061982791458521116MiniMax M3 × Fireworks: https://x.com/FireworksAI_HQ/status/2062187803476111405Brian Chesky's AI lab: https://x.com/shiringhaffary/status/2062618738881675579Is AI profitable?: https://isaiprofitable.com/v0 × Shopify: https://x.com/v0/status/2062859311869497355Factory Router: https://x.com/factoryai/status/2061862733126275549Hermes Desktop (Nous): https://x.com/NousResearch/status/2061843507417944552
⏱️ CHAPTERS00:00 Cold open00:36 Welcome — live from the CodeRabbit office00:56 Anthropic ships 8x more code (and the "depressed employees" reply)02:46 Bots pass humans: agentic traffic overtakes the web03:17 Loop engineering: stop prompting, start designing loops10:53 Subscribe break11:03 OpenAI Codex Sites — is Lovable cooked?11:56 Cognition's $10M AI Productivity Guarantee13:21 Devin Desktop (pour one out for Windsurf)14:03 Cloudflare acquires VoidZero14:16 The China shift: DeepSeek, Qwen, GLM, MiniMax15:39 Notion disables Anthropic models16:12 Model dump: Gemma 4, Magenta RT2, Miso One, Nemotron, Liquid, GLM 5.1, MiniMax M319:26 Funding & M&A: Suno, Supabase, SpaceX, Chesky's AI lab21:00 Is AI profitable?21:35 Quick hits: v0 × Shopify, Factory Router, Hermes Desktop22:45 Outro & thanks to CodeRabbit

WorkOS quietly powers the auth and enterprise layer under OpenAI, Anthropic, Cursor, and a long list of companies building the AI era. So when its founder decides his entire company should learn to code, it's worth asking why.
Michael Grinich joins Shane and Abhi to talk about what "AI-native" actually looks like from the inside — the in-house coding agent his team built instead of buying, the all-day internal hackathons, and the operating principle he now hires for: everybody codes.
He shares how AI is reshaping go-to-market and not just engineering, why he braced for pushback and got the opposite, and the one trait that separates the people who thrive from the people who stall. We close on the thinking behind his Death of UI talk, and where human-computer interaction goes from here.
🔗 MICHAELThe Death Of UI https://youtu.be/azoM7uz8JhsMichael Grinich on X — https://x.com/grinichWorkOS — https://workos.com
📚 MASTRA RESOURCEShttps://mastra.aihttps://x.com/mastra_aihttps://mastra.ai/community/discordhttps://github.com/mastra-aihttps://mastra.ai/coursehttps://mastra.ai/books/principles-of-building-ai-agentshttps://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRA?Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
CHAPTERS0:00 "Everybody codes"0:13 What WorkOS actually does1:33 Inside the Applied AI showcase2:39 Horizon, the in-house coding agent3:49 Build it or buy it5:50 WorkOS's marketing strategy7:29 Getting everyone building9:40 "We haven't found our limits yet"11:41 You have agency to learn13:38 Should every company build this?16:04 The Death of UI17:58 Where to find Michael

Anthropic filed to go public. The S-1 lands the same week as a $65 billion Series H at a $965 billion valuation, a claimed first profitable quarter, and a home listing that takes Anthropic stock as payment.
Opus 4.8 shipped mid-week, and Shane and Abhi give the vibe check: a polish pass over 4.7 more than a step change, still trailing GPT-5.5 on DeepSWE, with GPT-5.6 spotted in Codex logs. The quieter, more interesting release: mid-conversation system messages, steering a model mid-task without breaking the prompt cache.
MiniMax M3 is out, agent-tuned and long-context at a fraction of frontier pricing. NVIDIA used Computex to push AI onto the desktop with Vera, RTX Spark, and a Windows DGX Station that runs trillion-parameter models locally, plus a startup unit that bolts onto your house and pays you for compute.
OpenAI brings Codex to Windows and ships private MCP servers. Claude Code's dynamic workflows get cloned by pi within a day, Devin raises $1B, and the model-vs-harness debate heats up.
Plus continual learning as the next wave, a $500M accidental Claude bill, Corgi's seven-days-a-week firestorm, and a GitHub Star Party pick.
🔗 LINKS
Pope on AI: https://x.com/pontifex/status/2060322763718725798Kuzma on TBPN: https://x.com/tbpn/status/2060374399031632176Opus 4.8: https://x.com/alexalbert__/status/2060043196655362358Mid-conversation system messages: https://x.com/swyx/status/2060044644193624253Opus 4.8 on DeepSWE: https://x.com/arrakis_ai/status/2060757773579956640GPT-5.6 sighting: https://x.com/hqmank/status/2060334752369160472Anthropic S-1: https://x.com/AnthropicAI/status/2061478052257841495Series H: https://x.com/anthropicai/status/2060061347522433422First profitable quarter: https://techcrunch.com/2026/05/20/anthropic-says-its-about-to-have-its-first-profitable-quarter/$500M Claude bill: https://x.com/Polymarket/status/2060034216906068131Zillow listing: https://x.com/Yuchenj_UW/status/2060776120380010932Mythos class model: https://x.com/kimmonismus/status/2060047510853312557MiniMax M3: https://x.com/minimax_ai/status/2061266317815296322Computex in 12 minutes: https://youtu.be/ugNnw4lAMWANVIDIA Vera: https://x.com/nvidianewsroom/status/2061298380022726734RTX Spark: https://x.com/nvidia/status/2061313474005737829DGX Station for Windows: https://x.com/nvidianewsroom/status/2061307670607319201Home AI data center: https://x.com/w1nklerr/status/2060091525413884408Codex on Windows: https://x.com/OpenAI/status/2060428604727771421Private MCP servers: https://x.com/openaidevs/status/2059703536825565499Dynamic workflows: https://x.com/claudedevs/status/2060044853279617150pi dynamic workflows: https://x.com/micLivs/status/2060115468531499224Devin raises $1B: https://x.com/cognition/status/2059660758531940856grok-build: https://x.com/xai/status/2060392249402552457Codex as QA: https://x.com/steipete/status/2061208638027395490Cua on Windows: https://x.com/trycua/status/2059688960838828391Elad Gil on liftoff: https://x.com/eladgil/status/2061129428084887593Continual learning pivot: https://x.com/swyx/status/2061206120233054327Trajectory: https://trajectory.ai/Motion Studio: https://x.com/_adishj/status/2059666916835463646Koji AI tutor: https://x.com/suekhim/status/2060378988606878147Asana acquires Stack AI: https://x.com/techcrunch/status/2060091143556395421Corgi raises $106M: https://x.com/nico_laqua/status/2060028908704243782Coral board with Gemma: https://x.com/googlegemma/status/2059740184930074758Cloudflare web search: https://x.com/cherryjimbo/status/2060717359979958513Voice-controlled computer: https://x.com/farzatv/status/2060865350036750847rift: https://github.com/anomalyco/rift
📚 MASTRA RESOURCEShttps://mastra.aihttps://x.com/mastra_aihttps://mastra.ai/community/discordhttps://github.com/mastra-aihttps://mastra.ai/coursehttps://mastra.ai/books/principles-of-building-ai-agentshttps://mastra.ai/books/patterns-of-building-ai-agents
Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
0:00 Cold open0:29 Welcome + last week's callbacks2:22 Opus 4.85:51 Anthropic files to go public11:33 MiniMax M315:23 NVIDIA at Computex18:11 NVIDIA's home AI data center20:32 OpenAI ships for Windows21:15 More coding agent news25:07 Continual learning27:42 Quick hits34:04 GitHub Star Party

Shane and Abhi are back with AI news! 
Andrej Karpathy joined Anthropic. The OpenAI co-founder and former Tesla AI head said he wants to "get back to R&D" at the LLM frontier. Same week, Greg Brockman posted "the model alone is no longer the product."
Elon Musk announced SpaceX is offering AI compute as a service at significant scale, with Anthropic as the flagship customer. Tom Brown confirmed Anthropic is scaling on GB200 capacity in Colossus 2 through June. Anthropic is reportedly paying SpaceX $1.25B a month — a $15B run rate to one vendor.
OpenAI offered $2M in tokens to every YC startup in the current batch in exchange for equity. WorkOS launched auth.md, an open protocol for agents to register for services on the web, with Cloudflare and Firecrawl as launch partners.
The Chinese labs kept pushing. DeepSeek made their 75% discount permanent on V4-Pro. The architecture behind the price: V4's KV cache is 100x smaller, ~3GB VRAM for 1M tokens. Qwen shipped 3.7-Max. MiniMax teased a similar move.
Anthropic shipped self-hosted sandboxes and MCP tunnels for Managed Agents, /workflows in Claude Code that replaces the LLM orchestrator with code, /usage for per-component token attribution, and a first-party security plugin.
Google had a week. Gemini 3.5 Flash launched at 3x input and 6x output the price of 3 Flash — Theo's math shows it costs 2x more to run than 3.1 Pro on similar tasks. Gemini Omni lost a side-by-side to Seedance 2.0. Jack Wotherspoon shipped Antigravity CLI, Philipp Schmid debuted Managed Agents in the Gemini API, and Google open-sourced Agent Executor.
Cursor shipped Composer 2.5 and published CursorBench. Datacurve released DeepSWE as a harder agentic coding benchmark.
Supply-chain attacks kept rolling: Mini Shai-Hulud hit antv, TrapDoor crypto stealers spread across npm/PyPI/Crates.io, Megalodon injected 5,718 commits into 5,561 GitHub repos in six hours, and GitHub itself disclosed unauthorized access to its internal repositories. Cloudflare published their experience running Anthropic's Mythos against 50 of their own repos.
The MCP 2026-07-28 RC is stateless. ElevenLabs launched Music v2 and Speech Engine. Runway shipped Aleph 2.0. Accenture laid off 11,000 in an $865M AI restructuring. Exa raised $250M at $2.2B. OpenRouter raised $113M.
🔗 LINKShttps://x.com/karpathy/status/2056753169888334312https://x.com/gdb/status/2057670776803996110https://x.com/sama/status/2056933166875857290https://x.com/grinich/status/2057884407135187292https://x.com/deepseek_ai/status/2057854261699195173https://x.com/teortaxestex/status/2057728159479443927https://x.com/elonmusk/status/2057228707606196434https://x.com/pitdesi/status/2057207627567014014https://x.com/serenaa_ge/status/2059308218564890875https://cursor.com/evalshttps://x.com/claudeai/status/2056645485696315581https://x.com/theo/status/2056877869780107762https://x.com/JackWoth98/status/2056805210761077059https://x.com/_philschmid/status/2056836567470362955https://x.com/cursor_ai/status/2056415413077233983https://x.com/cloudflare/status/2056360412510060748https://x.com/dsp_/status/2057780712187580924https://x.com/elevenlabs/status/2059312414198235642https://x.com/runwayml/status/2057530497597600169https://x.com/exaailabs/status/2057132080317042697https://x.com/openrouter/status/2059277623629664758
📚 MASTRA RESOURCEShttps://mastra.aihttps://x.com/mastra_aihttps://mastra.ai/community/discordhttps://github.com/mastra-aihttps://mastra.ai/coursehttps://mastra.ai/books/principles-of-building-ai-agentshttps://mastra.ai/books/patterns-of-building-ai-agents
WHAT IS MASTRA? Mastra is an open-source TypeScript framework designed for building and shipping AI-powered applications and agents with minimal friction. It supports the full lifecycle of agent development — from prototype to production.
00:00 — Intro00:58 — The unlikely AI cast: Pope, Kuzma, Karpathy, Brockman05:36 — OpenAI offers $2M tokens to every YC startup06:40 — auth.md from WorkOS07:44 — China keeps shipping: DeepSeek, Qwen, MiniMax09:24 — Anthropic + SpaceX: $1.25B a month10:55 — Coding benchmarks: DeepSWE & CursorBench13:29 — Anthropic ships: sandboxes, /workflows, /usage, security15:08 — Google's week: Gemini 3.5 Flash, Antigravity, Managed Agents18:36 — Cursor Composer 2.519:03 — The supply-chain attack marathon22:31 — MCP 2026-07-28 RC goes stateless23:02 — Voice, music, video: ElevenLabs & Runway Aleph 2.024:52 — Accenture layoffs, Exa $250M, OpenRouter $113M26:24 — Quick hits