Stuff doesn't do itself | https://xitter.cf/t.co/iY7RVi1AC0
Germany
Joined September 2025
- Tweets749
- Following370
- Followers64
- Likes421
Cannot fathom this level of hatred and vitriol. This guy's entire life revolves around how much he hates OpenAI and especially Tibo
Tibo got community noted after FULLY denying the nerf ever happened and instead blamed users for fraud, but the screenshot of the note went viral anyway
So then he made another follow up post partially admitting it did happen, saying they’re “investigating”
But I suspect he added the word “BANKED resets” so people forget about the gaslighting and normies turbo like the post and make it viral
To maintain the illusion amongst the masses that OpenAI are the good guys who would never do such a thing
Narrative control
INTRODUCING Bowline.
You have your MacBook, your Mac Mini, and your Linux box all running agents, all editing code. Keeping them in sync is easy, just git push git pull, right?
WRONG, that's dumb and annoying. Bowline runs a tiny rust daemon on your machines and keeps your agent's code edits sync across devices almost instantly.
End to end encrypted, and yes your gitignored .env syncs too (no more manually copying API keys between machines).
Try it now at bowline.sh/
I choose to believe that the @TraycerAI sponsor in @theo's latest video is because of me
Same, I thought it was just me. I was excited to use Fable with it with GPT-5.5 but froze every time I tried it.
I've been using the @TraycerAI app, takes a little setup but the subagent abstraction with agent to agent mail is genuinely one of the best I've used.
Clipmem 0.6.0 is here!
• Better search with honest match scores
• Safer restores with rollback and no duplicate captures
• Consistent privacy rules across every capture path
• Source-safe image previews and durable OCR jobs
• Stronger deletion and database integrity checks
• Just in general better
Save all your clipboard history to a local SQLite db (redacting secret-looking-things if you choose). Images get OCR'd by a local model so they're searchable too, and there's a nice CLI for your agent and a nice MacOS GUI for you :)
The new @TraycerAI desktop app is legit, works very well for Fable orchestrating Codex/Opus sub-agents. Probably the smoothest experience with sub-agents I've had so far.
I think the big labs should release frontier models that haven't been deepfried with RL, for writing and talking to.
GPT-4.5 style but with modern architectures and more writing-focused training data rather than code.
I'm sick of being told my plans are unusually thorough, or that the model would not claim something, or that what I said is worth repeating because it's the load bearing part, and all the other weird LLMisms we get with the frontier models today.
Just talk like a human bro.
Guess who didn't realise they could only use up to 50% of their weekly limit for Fable and ran out 🤦♀️
Pro tip for releasing on Product Hunt: set up a Gmail filter to delete emails containing the word "backlink"
I just submitted CrowNest for review on Ship or Die 🏴☠️
It took me 6 days to ship in public and submit proof.
Follow the mission on my profile:
ship-or-die.com/u/tristan
