Links worth reading
Saved from the timeline, straight to the source ↗I was a Staff Engineer at Apple and built both internal APIs used by 100 as well as public iOS APIs. Here are some non-obvious tips for API design and maintainability 🧵 x.com
I was a Staff Engineer at Apple and built both internal APIs used by 100 as well as public iOS APIs. Here are some non-obvious tips for API design and maintainability 🧵
"Let’s plan the roadmap for next quarter” is the modern day equivalent of “let’s fax everyone the strategy.” We're going into q4 planning and it makes my brain hurt. Roadmaps existed because software x.com
A post by Jean-Michel Lemieux (jmwind) arguing that quarterly roadmaps are obsolete now that AI agents make building software cheap. He calls the replacement "Continuous Compilation": keep a list of ideas, write an ambitious one-page prompt for each (goal, use cases, rough architecture, success crit
e2e: The open source AI testing frameworkDiscordGitHubGitHubGitHub tester.army
Introducing e2e The agentic testing framework for any app ~/ npx e2e init → Fully Open Source → Mix deterministic and agentic APIs → Supports web, mobile, and more → Bring your own agent and infrastructure → Run locally or in CI
I literally can't think of a better risk/reward bet right now than building a ChatGPT plugin and getting in front of their 1.2 billion weekly users. As of yesterday, ChatGPT recommends plugins in the x.com
I literally can't think of a better risk/reward bet right now than building a ChatGPT plugin and getting in front of their 1.2 billion weekly users. As of yesterday, ChatGPT recommends plugins in the middle of conversations. Millions of questions get asked every day with no
The US Navy has ordered 30 medium unmanned surface vessels, 10 each from Saronic, HII and Galliano Marine Service, at an average of about 40 million dollars a hull, or roughly 1.2 billion dollars in t x.com
The US Navy has ordered 30 medium unmanned surface vessels, 10 each from Saronic, HII and Galliano Marine Service, at an average of about 40 million dollars a hull, or roughly 1.2 billion dollars in total. All of them will be built at shipyards in south Louisiana. Saronic's
a dad in Nashville open-sourced a full home security system that runs 100% locally. it's called Frigate, a AI-powered NVR that runs on your own hardware and does real-time object detection on any IP c x.com
a dad in Nashville open-sourced a full home security system that runs 100% locally. it's called Frigate, a AI-powered NVR that runs on your own hardware and does real-time object detection on any IP camera. → detects people, cars, animals, packages → 100+ detections per
This guy coaches Sam Altman. He also works with leaders at Anthropic, Google, Apple, and SpaceX. I sat down with him to learn how he coaches billion dollar businesses. Here are Joe Hudson’s 7 operatin x.com
This guy coaches Sam Altman. He also works with leaders at Anthropic, Google, Apple, and SpaceX. I sat down with him to learn how he coaches billion dollar businesses. Here are Joe Hudson’s 7 operating rules: 1. Kill your 1:1s These meetings lead individuals to solve problems
ex-Apple dev with 25 years in cybersecurity showed me his entire agentic engineering workflow x.com
David Ondrej interviews Daniel, an ex-Apple security veteran, about agentic engineering and AI security. Daniel argues the future is attacker AI vs defender AI, so every company needs its own model stack and harness, ideally with internal hardware and open-source models to avoid lab dependency. He w
dotfiles/skills/optimise-github-actions/SKILL.md at main · enesgules/dotfiles · GitHub github.com
our github actions was using 43k minutes a month 💀 most of it was waste so i made a skill that finds this stuff for you 𝚗𝚙𝚡 𝚜𝚔𝚒𝚕𝚕𝚜 𝚊𝚍𝚍 𝚎𝚗𝚎𝚜𝚐𝚞𝚕𝚎𝚜/𝚍𝚘𝚝𝚏𝚒𝚕𝚎𝚜 --𝚜𝚔𝚒𝚕𝚕 𝚘𝚙𝚝𝚒𝚖𝚒𝚜𝚎-𝚐𝚒𝚝𝚑𝚞𝚋-𝚊𝚌𝚝𝚒𝚘𝚗𝚜
Grok Bot will find you $100,000 at 0% interest in about 5 minutes if you paste this in "You're my business funding analyst. My credit score is [score]. I live in [state]. I have an LLC and EIN: [yes/n x.com
Grok Bot will find you $100,000 at 0% interest in about 5 minutes if you paste this in "You're my business funding analyst. My credit score is [score]. I live in [state]. I have an LLC and EIN: [yes/no]. My personal cards are [bank, limit, balance for each] 1. Find every US
via hunter @hxxntrr
Everybody wondering why Omarchy is succeeding should read this classic Joel on Software essay: joelonsoftware.com
Everybody wondering why Omarchy is succeeding should read this classic Joel on Software essay:
Founders: Your LinkedIn-and-Slack GTM dies when the buyer is on a job site, not in a browser tab. I recorded a Main Street GTM class for Founder Led Sales Factory students who sell that way. Construct x.com
Founders: Your LinkedIn-and-Slack GTM dies when the buyer is on a job site, not in a browser tab. I recorded a Main Street GTM class for Founder Led Sales Factory students who sell that way. Construction, home services, industrial, solar installers. People who do business face
SpaceXAI engineer (ex-Cursor): "Less than 1% of the world has ever touched Cursor, Claude Code or Codex. That gap is the biggest opportunity of your career I write 0% of my code by hand now. Agents sh x.com
SpaceXAI engineer (ex-Cursor): "Less than 1% of the world has ever touched Cursor, Claude Code or Codex. That gap is the biggest opportunity of your career I write 0% of my code by hand now. Agents ship it, I decide what gets built" In 76 minutes, engineers from SpaceXAI,
Hearing this week that personal assistants cost $3k-$7k per user per year, depending on scale effects / vertical integration / model selection. Implications/questions: - Consumer subsidies will be a h x.com
Hearing this week that personal assistants cost $3k-$7k per user per year, depending on scale effects / vertical integration / model selection. Implications/questions: - Consumer subsidies will be a huge factor; people don’t care about privacy, they want utility and they want it
Removed 60,984 lines of test slop :) x.com
Removed 60,984 lines of test slop :)
OpenClaw deleted around 400k LOC of its own tests without much change in code coverage. Modern models just love writing tests for every tiny change, even if they aren't useful. This skill helped. github.com
OpenClaw deleted around 400k LOC of its own tests without much change in code coverage. Modern models just love writing tests for every tiny change, even if they aren't useful. This skill helped.
Yup. x.com
Yup.
I find the latest models write a lot of unit tests. Most of them restate the code they were written after, so they always pass and catch almost nothing. They run fast, but they break on every refactor x.com
I find the latest models write a lot of unit tests. Most of them restate the code they were written after, so they always pass and catch almost nothing. They run fast, but they break on every refactor and the agent burns time fixing tests instead of the feature. The prompt I
Introducing Rollouts. Rollouts write a monitoring plan, then watch changes as they deploy. Deployments are verified, so regressions are caught before users see them. x.com
Introducing Rollouts. Rollouts write a monitoring plan, then watch changes as they deploy. Deployments are verified, so regressions are caught before users see them.
The first time I used a terminal was when I opened Cygwin in CS101 and typed pwd. Since then, I’ve used a terminal as one of my primary work tools every day. But it wasn’t until coding agents that I w x.com
The first time I used a terminal was when I opened Cygwin in CS101 and typed pwd. Since then, I’ve used a terminal as one of my primary work tools every day. But it wasn’t until coding agents that I wanted to change the terminal itself. I’m really excited to share my side
Price discovery for your career / go to market for getting an amazing job. If you're a swe in sf/nyc there are hundreds of startups that want to hire you. And if you're lit you already know there are x.com
Price discovery for your career / go to market for getting an amazing job. If you're a swe in sf/nyc there are hundreds of startups that want to hire you. And if you're lit you already know there are people hitting you up but it's across your dms, inbox, linkedin and more. When
Stanford CS336 is crazy bro , You think you too behind in the AI/ML p try CS336. You go from Zero to one in oneplaylist maths is a prerequisite tho. x.com
Stanford CS336 is crazy bro , You think you too behind in the AI/ML p try CS336. You go from Zero to one in oneplaylist maths is a prerequisite tho.
who's doing agent phone calls with their bots and what' the best platform thats like 1-2 levels above twilio to make natural sounding voice calls to businesses etc where you get the full log /transcri x.com
who's doing agent phone calls with their bots and what' the best platform thats like 1-2 levels above twilio to make natural sounding voice calls to businesses etc where you get the full log /transcript PLUS you can jump in by typing something while the agent is on a call
one-shot /animate skill from @emilkowalski x.com
one-shot /animate skill from @emilkowalski
via Andy @andywang
if I was Dumbledore I would've hid the philosopher's stone in the aws console x.com
if I was Dumbledore I would've hid the philosopher's stone in the aws console
I made a prompt that checks any AI answer before you act on it using the CIA's Competing Hypotheses method. Richards Heuer built it for CIA analysts in the 1970s around one habit. Don't collect eviden x.com
I made a prompt that checks any AI answer before you act on it using the CIA's Competing Hypotheses method. Richards Heuer built it for CIA analysts in the 1970s around one habit. Don't collect evidence for the answer you like. List every explanation that could be true and try
We assume that agents need separate context windows for separate tasks, and that these context windows should not mix. But is that correct for a future where agents are much smarter than humans? We ga x.com
We assume that agents need separate context windows for separate tasks, and that these context windows should not mix. But is that correct for a future where agents are much smarter than humans? We gave our cloud agents their own computers with exactly 1 running context window
Read this. Read all of it. And then draw your own conclusion. My conclusion: he’s right on. x.com
Read this. Read all of it. And then draw your own conclusion. My conclusion: he’s right on.
This is the way. Accept code on merit and ensure that a person takes accountability for it. Doesn’t matter if it was typed, chiseled, generated, or bit-flipped via magnetized needle on a chip. If it’s x.com
This is the way. Accept code on merit and ensure that a person takes accountability for it. Doesn’t matter if it was typed, chiseled, generated, or bit-flipped via magnetized needle on a chip. If it’s good, it’s good. If it’s slop, it’s slop.
via tobi lutke @tobi
Steal these musecases: 10 prompts that turn Muse from a chatbot into your personal operator. 1. Money: "Audit my last 3 months of subscriptions and tell me what to cancel, with amounts." 2. Bills: "Ch x.com
Steal these musecases: 10 prompts that turn Muse from a chatbot into your personal operator. 1. Money: "Audit my last 3 months of subscriptions and tell me what to cancel, with amounts." 2. Bills: "Check every recurring bill against current market rates and flag overpayments."
Codex 接入 Jev 后，我的token额度消耗节省了90% 我实测输入一句： “游戏机我要买吗？” Jev 很快判断为 “购买咨询”，概率 97%，置信度 0.96，然后直接进入后续处理 它最适合干这些事，分类、筛选、路由、快速判断 Codex 则继续负责复杂推理。 接入也非常简单： 先去 typesafe.ai
Codex 接入 Jev 后，我的token额度消耗节省了90% 我实测输入一句： “游戏机我要买吗？” Jev 很快判断为 “购买咨询”，概率 97%，置信度 0.96，然后直接进入后续处理 它最适合干这些事，分类、筛选、路由、快速判断 Codex 则继续负责复杂推理。 接入也非常简单： 先去
最近做 APP 一直在用一个 Skill 叫 Apple Design，将 Apple WWDC 设计分享中的界面与动效原则整理成指导，帮助改善界面层级、反馈和整体连贯性。实践效果非常好，简单克制、很有苹果原生 App 的味道。 作者Emil Kowalski 之前在 Vercel 现在是 Linear 的设计师，Apple Design 是他整理的一组 AI Agent x.com
最近做 APP 一直在用一个 Skill 叫 Apple Design，将 Apple WWDC 设计分享中的界面与动效原则整理成指导，帮助改善界面层级、反馈和整体连贯性。实践效果非常好，简单克制、很有苹果原生 App 的味道。 作者Emil Kowalski 之前在 Vercel 现在是 Linear 的设计师，Apple Design 是他整理的一组 AI Agent
via 酱紫表 @pengchujin
I think a very fun idea is to have your @muse maintain a Spotify playlist for you that it is constantly adding and removing songs from. Kind of like a bucket of 25 songs that you can listen to through x.com
I think a very fun idea is to have your @muse maintain a Spotify playlist for you that it is constantly adding and removing songs from. Kind of like a bucket of 25 songs that you can listen to throughout your day. Kind of like your own personal DJ. Your muse can even track
介绍比Jev快50倍，在你设备上跑的laya-mlx! 只在你的设备上占用最高1G内存 Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统 我将其移植到MLX,并且做了一些性能优化! 视频中就是这个模型在我的本地M3Max上玩贪吃蛇 这个模型能够以每秒决策60次的速度玩贪吃蛇！ x.com
介绍比Jev快50倍，在你设备上跑的laya-mlx! 只在你的设备上占用最高1G内存 Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统 我将其移植到MLX,并且做了一些性能优化! 视频中就是这个模型在我的本地M3Max上玩贪吃蛇 这个模型能够以每秒决策60次的速度玩贪吃蛇！
via 雨夹雪❄️ @mizorewww
It’s been 2 hours and I can already tell this is going to be my most popular post. See the replies. And I’m getting so many DMs from people who went through the same, I’m glad it’s connecting. 2 quick x.com
It’s been 2 hours and I can already tell this is going to be my most popular post. See the replies. And I’m getting so many DMs from people who went through the same, I’m glad it’s connecting. 2 quick things - - please consider the solution!! it’s counterintuitive and no one
I can’t emphasize how important it is to surround yourself with people who can give you high quality advice like this. You need friends to stop you from doing something stupid ideally BEFORE you do it sunilpai.dev
I can’t emphasize how important it is to surround yourself with people who can give you high quality advice like this. You need friends to stop you from doing something stupid ideally BEFORE you do it. Amazing advice @threepointone
via Keegan @keegabit
Opening access for developers to build Muse connectors. You bring the API -- Muse brings the agent, the browser, and the context of what the person actually wants. People reach your service just by as x.com
Opening access for developers to build Muse connectors. You bring the API -- Muse brings the agent, the browser, and the context of what the person actually wants. People reach your service just by asking for it, and their agent takes it from there. New connectors are live
Founders: Founder-Led Sales 101, from my lecture at Harvard Business School. Free. No signup. Just the talk: drive.google.com
Founders: Founder-Led Sales 101, from my lecture at Harvard Business School. Free. No signup. Just the talk:
Given that agents still need high level supervision. And given that the human review of code is a bottleneck. And given that the models have gotten good enough that strict deterministic harnesses are x.com
Given that agents still need high level supervision. And given that the human review of code is a bottleneck. And given that the models have gotten good enough that strict deterministic harnesses are suboptimal. Then how should we supervise their output. My current solution
got Jev to review my PRs. ~200x cheaper than Claude and it answers in half a second 6 real PRs in the video. $0.00007 each. 1,000 PRs = 7 cents vs ~$14.50 on Opus 5 paste a diff → ONE call to @typesaf x.com
got Jev to review my PRs. ~200x cheaper than Claude and it answers in half a second 6 real PRs in the video. $0.00007 each. 1,000 PRs = 7 cents vs ~$14.50 on Opus 5 paste a diff → ONE call to @typesafeai → 14 typed checks come back as probabilities: hardcoded secret, sql
A trucking company owner trained a bot on his own business in eight days and posted the list of what it does now. Twelve numbered jobs. Dispatch questions, driver paperwork, invoice chasing, the lot. x.com
A trucking company owner trained a bot on his own business in eight days and posted the list of what it does now. Twelve numbered jobs. Dispatch questions, driver paperwork, invoice chasing, the lot. Not a tech company. Trucks. The tools stopped being for developers around
this absolute stud saw my video on how to rank in AI Search in 24 hours he followed my advice now he's ranking in AI Search (in <24 hours) if you want rank and get leads like Axel: use this exact SOP inboundengineer.io
this absolute stud saw my video on how to rank in AI Search in 24 hours he followed my advice now he's ranking in AI Search (in <24 hours) if you want rank and get leads like Axel: use this exact SOP (this is what I sent him/got him ranking) 👇
i need you to understand what is happening in marketing engineering right now i want to try to explain this to you if you give your claude code or codex: - Data pipeline - Data warehouse - Cloud Serve x.com
i need you to understand what is happening in marketing engineering right now i want to try to explain this to you if you give your claude code or codex: - Data pipeline - Data warehouse - Cloud Server - Media Storage (Images, Videos) - Databases for agents (Postgres
let's rename vibe coding to cuck coding cause you just sit there and watch occasionally giving direction x.com
let's rename vibe coding to cuck coding cause you just sit there and watch occasionally giving direction
via Coins @Coins
A 118K-star GitHub repo is built around one painfully obvious idea: Your coding agent should map the codebase once, not grep it forever. Graphify turns an entire project into a queryable knowledge gra x.com
A 118K-star GitHub repo is built around one painfully obvious idea: Your coding agent should map the codebase once, not grep it forever. Graphify turns an entire project into a queryable knowledge graph. Functions, classes, files, SQL schemas, infrastructure, docs, PDFs,
The software factory approach is getting pretty popular, but it can be daunting to adopt all at once. Let's go through “crawl, walk, run” steps for making the transition from local, interactive agents x.com
The software factory approach is getting pretty popular, but it can be daunting to adopt all at once. Let's go through “crawl, walk, run” steps for making the transition from local, interactive agents to automated cloud development.
I am once again telling you that a coding agent can run your google ads account better than a human media buyer give it data for SEO api, google ads api, all reads from a data warehouse so you don't h x.com
I am once again telling you that a coding agent can run your google ads account better than a human media buyer give it data for SEO api, google ads api, all reads from a data warehouse so you don't hit api rate limits, then let it write software to run the account daily have a
Interesting to see Yegge say he never successfully built anything with Gas Town. In I mentioned not finding these ultra vibed orchestrators useful b/c reliability (w.r.t. completing tasks). Turns out danluu.com
Interesting to see Yegge say he never successfully built anything with Gas Town. In I mentioned not finding these ultra vibed orchestrators useful b/c reliability (w.r.t. completing tasks). Turns out the author of the most famous one had the same issue.
via Dan Luu @danluu
The CRAP (Change Risk Anti-Patterns) metric is a software quality measurement that combines code complexity and automated test coverage to calculate the risk of modifying a function or method. CRAP(m) testing.googleblog.com
The CRAP (Change Risk Anti-Patterns) metric is a software quality measurement that combines code complexity and automated test coverage to calculate the risk of modifying a function or method. CRAP(m) = comp(m)^2 * (1 - cov(m))^3 + comp(m)
via Stefan @schteppe
Prompt I use a lot: "how can I pokayoke this so this kind of error never happens again" pokayoke is using simple, low-complexity methods to make a particular kind of mistake impossible to make. classi x.com
Prompt I use a lot: "how can I pokayoke this so this kind of error never happens again" pokayoke is using simple, low-complexity methods to make a particular kind of mistake impossible to make. classic one would be connector shape: make it impossible to insert "the wrong way".