Staff Software Engineer exploring all things Gamedav, Graphics, and Spatial computing Previously: Mozilla Hubs, AltspaceVR
New York, NY
Joined January 2008
- Tweets2.8K
- Following943
- Followers590
- Likes3.4K
This sucks. Not because I want to clone games, but anyone who has worked on games knows that you constantly take inspiration from existing things and adapt it in your own way to your own work.
And anyone who has worked with agents knows that these guardrails constantly false positive.
So now working on a game is going to be a constant annoying tiptoe around mentioning anything even approaching existing games.
ANTHROPIC IS CRACKING DOWN ON VIDEO GAME REWRITES 🚨
Claude Code picked up a new rule around Oct 27, sent from Anthropic’s servers under a flag called "tengu_heron_brook". There’s no setting to turn it off.
The full rule: “Do build original games, games in a genre, and games from the user’s own designs and assets. Don’t build a recreation of a specific existing game, even if it is renamed or called original.”
Is this the end of game rewrites, modding and everything in between?
Ugh. Day 3 waiting on appstore review. This sucks. I already have the next update ready to go. I really hope it speeds up after the launch. And praying I don't get kicked back a rejection after all this time over something dumb.
Why do OpenAI and Anthropic keep working on their own desktop and mobile apps when @t3dotcodes exists. It's open source, just throw them some cash so they can put more devs on it, whitelabel it if you must, and call it a day.
Focus on making kick ass models and infra.
It's really fascinating that Claude is able to understand Codex speaking to it in grugspeak. Like I wonder if they actually train on this or it just smart enough to put all the context clues together.
I can parse it if I really try when I have all of the context but it's rough.
I may need to move up investigating git-cow and simslim. I keep blowing through my disk use with worktrees and iOS simulator instances. Easily bursting up and down like 150gb+ at a time.
Haven't tried it yet but I wonder how much something like git-cow would help. If nothing else will save a lot of disk space.
github.com/gabrielkoerich/gi…
I am migrating years of Google Maps stars and favorites over to Apple Maps to finally give it a real go since Google Maps kind of sucks on the iPhone Duo right now. Terra is doing a great job with it.
Hey @FrameworkPuter. So I ordered a refurb framework desktop 26 days ago that said (and still says) it would ship in 5 days.
After a few weeks of back and forth with support I was told it was backordered (even though it was not when I ordered it). In fact, when I first placed the order I tried to add an fan to the order minutes later and support told me it was not possible because the order was already "transmitted for processing and shipment".
Since having ordered it, the non-refurbished 64gb desktops have come back in stock. I want to know if my original order is ever going to ship but have now been waiting 5 days for a response from support. Afraid they are going to sell out again before I get any answers, so I am begrudgingly doing the beg on X thing.
Love your mission, love your stuff, but this has been a pretty poor order experience.
I realized now that I am effectively a PM that I was right all along as an IC thinking kanban, jira, and other PM slop are useless. Why the hell would I want to work like that. "Just care a lot about the product & talk to the <entities> doing the work" is still the right strat.
Don't think we are there yet, but something like this is what is going to be the viral breakout that starts the "AI Personhood" tribal war.
What the fuck. Gave it a link to the repo for this and the track Virtual Insanity. Only content direction I gave it was "maybe use the Claude character and give it the Jamiroquai hat. The music video is pretty well known so any nods to that would be kind of funny/interesting"
Expecting the agent would do anything here but answer the question is insane. It does make some sense to me why someone who expected the agent to infer that for some reason and act on that question without asking would prefer Fable over Sol. The later thing in the thread where it didn't babysit the followup is a bit more reasonable to get mad about, but it looks like it just didn't wait long enough for review comments to come in.
Some notes on this:
1. A lot of my problems were addressed by a newer snapshot we got later in testing
2. More of them were also addressed by deleting more of my agentmd
3. You can see the most cursed agent thread I’ve had this year here: 6a7ypw1fx9le.postplan.dev/