Carter Leffen

@SebAaltonen This Everything needs to be micromanaged The only way to keep this thing on rails is to force it to run an unshakable performance test every diff it makes Amusingly, when you do that, it knows it will get caught lying so it becomes a better programmer..

like the energizer bunny, /goal keeps on going and going

@yacineMTB gpt-5.5 has thumbs!?

guess who has two thumbs and is decompiling binaries

@seelffff obv the perfect stack for a pre-launch pre-user saas startup

ex-Googlers published a map of every internal tool Google uses and its open-source equivalent. 15,200 stars. 1,100 forks. 99 contributors. → Borg = Kubernetes → Spanner = CockroachDB → Colossus = HDFS → Dremel = DuckDB / Presto → Chubby = Zookeeper → Stubby = gRPC → Zanzibar = SpiceDB → Blaze = Bazel → MapReduce = Spark everything Google engineers use every day. all of it has an open-source equivalent. none of it requires working at Google. like+bookmark

@KSimback if you put “LK-99” at the beginning does it try to superconduct?

I thought I knew Claude well, but I just today learned there are some shorthand “secret” codes 1. Put “L99” at the end of a prompt and it makes Claude respond at “Level 99” maximum expert depth and reduces hedging and forces committed recommendations 2. Put “OODA” at the beginning of a prompt and Claude will structure its response using the military OODA loop framework (Observe -> Orient -> Decide -> Act) 3. Put “SCAFFOLD” at the beginning of a prompt for Claude to generate a full step-by-step project setup or action plan There are a few others but those 3 I found most useful These are not official commands but apparently they work because Claude’s training data contains thousands of examples of people using them So the model has learned the patterns and adjusts its response style accordingly Pretty cool to know, if you’ve found other good ones drop them in the comments

team openai has been posting from a position of confidence i’ve not seen in a long time

Principles of good software architecture Make your systems honest about what they are, ignorant of what they don't need to know, easy to verify, and cheap to change. Respect the inevitability of change. Pay attention to the parts that are likely to change, hide them behind stable abstractions, let the rest be simple. If it can't be safely modified it's a liability, not an asset. Keep change cheap: cheap to understand, cheap to verify, cheap to make, cheap to deploy. Ship in small steps, with feedback loops. Treat your code, tests, hooks, and agents as one system together.

my daughter knows how to melt my heart

Submitting RLM-FORGE to the @NousResearch Hermes Hackathon. RLMs were supposed to live inside inference. RLM-FORGE pulls them out into runtime. Hermes gives that runtime bounded calls, tools, skills, provider portability, and memory. Ouroboros gives it recursion, retry, state, and replay. TraceGuard makes sure parent claims cannot commit without fresh child evidence. The weird part is the memory: not as evidence, but as an operational prior for where to recurse next. RLM escaped inference. Hermes gave it memory. Evidence before belief. Repo: github.com/Q00/rlm-forge

@b_nnett bro you are cooking

Edit files directly within Codex, now available.

@tuturetom keep up the great work

Open Design 正在从一个简单的 Claude Design 的开源复刻转向真正的一站式开源 Design Studio ⚡️ 我们希望在 26 年的今天，借助最强的 Coding Agent，能够让最前沿的设计在一处完成，让最优质的设计触手可及 人人都是设计师的时代可能即将到来💪 关注我们 👉 github.com/nexu-io/open-d…

@TimJayas the someone is @tuturetom

CLAUDE CODE DESIGN BUT IT'S FREE! Someone cloned Claude Code and make it completely free > Generates UI/UX, dashboards, slides1 > Local first open source clone of Claude Design > Plugs into Claude, Cursor, Codex, Gemini CLI, OpenCode, Ollama, etc 100% Open Source and Free 👀

@gregpr07 codex for mac is awesome

Who actually uses Codex over Claude Code? Claude Code is just 100x better imo, like the DX is WAY better.

advertisements come fast

@pmarca @ctjlewis too bad not groucho marx

Co-sign. Most genius SF AI engineers derive their economics from a combination of Marx and science fiction (but I repeat myself).

@thsottiaux i can transfer a goal from cli to the cloud and back?

You can now keep codex going for days. With GPT-5.5 it will build an entire OS kernel for you if you ask, or find critical bugs in a codebase, or optimize your database schemas, or… the options are endless.

@sachintwtss 1) Ask it to install Codex 2) ??? 3) Profit

+ RLMs!