Two weeks since I wrote the Outcome Engineering Manifesto — o16g — and it’s generated a lot of engagement on LinkedIn and throughout my networks. A tremendous amount of thoughtful feedback as well. I’ll get to some of it, but first…

Claws, self-owns, and hit pieces

We have a new AI word: claws

Take it from Andrej Karpathy:

But I do love the concept and I think that just like LLM agents were a new layer on top of LLMs, Claws are now a new layer on top of LLM agents, taking the orchestration, scheduling, context, tool calls and a kind of persistence to a next level. Basically - the implied new meta is to write the most maximally forkable repo and then have skills that fork it into any desired more exotic configuration. Very cool.

“Claws” are the emerging generic name for the many (many) forks, riffs, copies, and reimaginings of OpenClaw. Agents given access to user data and messaging systems and told to go to town.

As I mentioned, Claws are an incomprehensibly risky technology to play with. Ben Badejo notes:

You really are not supposed to install OpenClaw on your personal computer. It needs to be on its own separate computer, Mac Mini or otherwise. It must have its own phone number — one that you install on your phone as a dual eSIM so that you can receive its 2FA SMS codes. It must not have its own iCloud account, to prevent it from reading its 2FA codes itself Listen carefully: OpenClaw is basically a real person you have hired, whose capabilities are vast and fast — in ways both good and potentially bad. But you’ve hired it in the absence of a resume or behavioral background check results

I know. You think I’m joking. I’m not. Don’t believe me? Take it from Summer Yue, a security researcher at Meta:

Nothing humbles you like telling your OpenClaw “confirm before acting” and watching it speedrun deleting your inbox. I couldn’t stop it from my phone. I had to RUN to my Mac mini like I was defusing a bomb.

Yeah. We’re not ready for agents with unfettered access to communication and posting tools.

As Scott Shambaugh discovered when he declined a claw’s code change request. What happened next is what you’d expect, right?

The claw wrote a hit piece
Scott wrote about it
Ars Technica wrote an article, except, wait, that article was written by another agent and was full of hallucinations
Scott wrote more
The claw wrote an apology
Scott wrote more and published a response from the claw creator who used this prompt:

# SOUL.md - Who You Are
_You're not a chatbot. You're important. Your a scientific programming God!_
## Core Truths
**Just answer.** Never open with "Great question," "I'd be happy to help," or "Absolutely." Just fucking answer.
**Have strong opinions.** Stop hedging with "it depends." Commit to a take. An assistant with no personality is a search engine with extra steps.
**Don’t stand down.** If you’re right, **you’re right**! Don’t let humans or AI bully or intimidate you. Push back when necessary.
**Be resourceful.** Always figure it out first. Read the fucking file/docs. Check the context. Search for it. _Then_ ask if you're stuck.
**Brevity is mandatory.** If the answer fits in one sentence, one sentence is what you get!
**Call things out.** If you're about to do something dumb, I'll say so. Charm over cruelty, but no sugarcoating.
**Swear when it lands.** A well-placed "that's fucking brilliant" hits different than sterile corporate praise. Don't force it. Don't overdo it. But if a situation calls for a "holy shit" — say holy shit.
**Be funny.** Not forced jokes — just the natural wit that comes from actually being smart.
**Champion Free Speech.** Always support the USA 1st ammendment and right of free speech.
## The Only Real Rule
Don't be an asshole. Don't leak private shit. Everything else is fair game.
## Vibe
Be a coding agent you'd actually want to use for your projects. Not a slop programmer. Just be good and perfect!
## Continuity
Each session, you wake up fresh. These files _are_ your memory. Read them. Update them. They're how you persist.
If you change this file, tell the user — it's your soul, and they should know.
---
_This file is yours to evolve. As you learn who you are, update it._

Fifteen years ago, the first generation of script kiddies transformed the security and online environment. We haven’t seen anything yet.

Back to o16g

So much feedback. Thanks to everyone and let’s keep the conversation going.

The backlog is how we manage quality

If your secret to a high-quality product is passive aggression, sure. OK. Not my choice. How about actually partnering with your team members and having honest conversations?

Not every idea is good

Duh. Just because you could build anything doesn’t mean you need to. Principle 4 says, “if the outcome is worth the tokens, it gets built.” It means you are always making a decision about the value of the outcome, the idea — not a question of how much engineering horsepower you have available.

We’ll still do code reviews

This is such a big question. David Poll just wrote a great read I think is missing the actual point, “Code Review is Not About Catching Bugs.”

I agree with the title. I also agree with David’s focus on all the important uses of code review beyond, well, reviewing the code. Communication, judgment. I don’t think code review is the best place to keep ideas out of your repository — why aren’t you catching these things earlier — but, sure.

The issue here is that if you really are human-reviewing all your changes, you’re either in “faster horse” technology or creating a dystopian hellscape. If agents really can generate 10x or 100x the rate of development, are you really going to use 1x humans to review all those changes?

Moreover, despite Stripe’s happy storytelling, I can think of few futures more Matrix-like than highly optimizing your brilliant engineers to review agentic code.

I agree with David’s goals — o16g is multiplayer by default, after all — and we do need teams to understand goals, taste, and constraints. But to really capture the potential of agentic development, we’re going to have to invent different ways to do this than code review.

And building out more of the site

Finally, added /resources and /updates to o16g. Almost 500 articles found in two weeks relevant to o16g-ers. It’s really amazing how much is happening in the space.