📦 EVERY ENTRANT GETS A FEEDBACK FILE 📦 📄 Your file is here: https://www.skool.com/cliefnotes/classroom/7634b927?md=2b3ccc0b73c243b7b6137993f6b60da3 🔍 HOW WE READ THESE We cloned every repo and pinned it to its last commit before the deadline. Nobody committed after the clock. Seven of seven. First time. Then we did what the rubric asked, in order, for every entry. We checked original/ against the judged commit, byte for byte. We read your feedback file first and wrote your three lines down before we opened anything else. We diffed that file against the one we sent you. We ran git log on original/ and on questions.md against rebuild/. We mapped every question to a line. Then we ran both arms ourselves, old and new, on your own harness where it would run on our machine and on ours where it would not. Then we wrote one question you had never seen, aimed at one of your three lines, with the expected answer written down before we ran it, and ran it on both arms. We read results.md last. That question is in your file, with what each arm did. 📚 WHAT THE FIELD TAUGHT ✅ Every entry closed at least two of its three lines. Four closed the tier line outright. The feedback files were specific enough to act on, and this is the first round where we can say that with evidence behind it: on our machine, on a question the author never saw, the rebuild beat the original on the line it was aimed at. ✅ The order held in git and slipped in the work. Six of seven committed questions.md before rebuild/. Three of those say in their own results.md that the rebuild was finished, or the questions shaped on a prototype of it, before the questions were written. Every one of them said so plainly. The step the brief asked for is about when the questions were written, and git can only show when they were committed. The one question nobody could tune to was ours. ✅ The harness is part of the entry. One rebuild lost on paper and won on our machine, because its system prompt was cut at the first quoted string before it reached the model. Another prints the right stop shape every time and fails its own checker on the prose the session writes around it. A third's validator needs a version of Python the original never did. What you measured is what you shipped.

A critical bug in the Linux kernel was discovered. It's called CVE-2026-98432. It is rated 9.8 out of 10. Anthropic just showed that it can be used to take over a server remotely. If you have a VPS, you are most likely affected. This includes Hostinger. Ubuntu has not released a fix yet. Paste this into your agent right now: "SSH into my VPS and check if it is vulnerable to CVE-2026-98432 (https://ubuntu.com/security/CVE-2026-98432). If a patched kernel is available, install it and reboot. If not, apply the safest temporary protection and tell me what you changed." Run it again when Ubuntu releases the fix.

Congratulations to @Curtis Hays ! Curtis is one our OG's and he takes the lead this week! He already has Lifetime VIP, so this time he gets a 30 minute 1-on-1 with Jake. ⏰ The 7-day clock resets. Next Monday someone new takes the top spot. 🎯 How it works: - 📝 Post things worth reading - 💬 Help people in the comments - 🛠️ Share what you're building, what's working, what's breaking - ❤️ Engage with what others are putting out there The leaderboard tracks all of it. 🎁 The prize, depending on where you're at: 🆓 Free member? You get lifetime Premium, free ⭐ Already Premium? We convert your Premium so you stop paying 👑 Already VIP? We convert your VIP so you stop paying 💎 Lifetime VIP? Get a 30 minute 1-on-1 with Jake. Either way, you stop paying.

Hi everyone! I had to share my latest project. I am so proud of it and hope you enjoy it! Thanks to @Jake Van Clief and his crew for creating the community. I feel like nothing is impossible to create. I had Claude code write the next part: I gave Claude one sentence. Two days later it was a 9.5 minute animated film. My idea was one line: "a child's room where different art styles are used." What came out is "The paper plane": a 9 minute 31 second animated film with 25 art styles, 8 narrator voices and an original soundtrack. Not one image or video generator was used. Every frame is drawn in code. Here is how it came together, and why ICM is the reason it worked. THE IDEA I had a folder of 84 reference art styles (sumi-e ink, claymation, film noir, Win95, synthwave and so on) and one picture in my head: a child's room where every object is a door into another style. The camera dives into the TV and you are inside a video game. The poster on the wall turns out to be a whole world. One style draws an arcade machine, and inside it the next style is already playing. THE BRAINSTORM Claude ran a structured brainstorm with me (the superpowers brainstorming skill): one question at a time, and a quick mockup of the room so I could see the camera route. Out of that came four "hand-off moves" that join one style to the next: - Dive in: through a screen into its world. - Drawn doorway: my arcade idea. - Pull back out: the world was the poster on the wall all along. - Carried thread: something crosses the seam from one style to the next. Then Claude pitched me three stories: 1. The paper plane: a plane that flies from style to style. 2. The bedtime question: a child asks something, and the room answers. 3. How pictures learned to move: from shadow puppets to 3D games. I loved all three and could not pick one. So I asked Claude to weave them into one story, and to drop the 2 to 3 minute limit we had planned, because a great story needs time to play out.

I'm building ATLAS, my personal multi-agent AI system, and this four-panel comic captures one of the biggest lessons I've learned. Panel 1 — What's missing? An AI can produce a brilliant answer while missing the original evidence. Panel 2 — Confidence ≠ Truth A polished answer isn't proof. Neither is agreement between multiple AI agents. Panel 3 — Receipts Matter I'm developing an evidence system built around original files, cryptographic hashes, verification, and traceable claims. Panel 4 — STOP. Find the originals. Before ATLAS makes a consequential claim, it needs to know whether the evidence actually exists. My two computers, Monster and Ghost, are potential sources—not automatic proof. The engineering principle behind it: PROPOSED → BUILT → TESTED → VERIFIED Those are four different states. An AI shouldn't get to skip any of them just because it sounds convincing. I'm learning that building reliable AI isn't just about making agents smarter. It's about making them accountable. Intelligence is capability. Evidence earns trust. Curious how others in Clief Notes are handling verification, agent memory, and hallucinations in their builds. #AIAgents #AIEngineering #BuildInPublic #ATLAS 👻