Max Milne
banner
maxmilne.bsky.social
Max Milne
@maxmilne.bsky.social
Psychotherapist.
All my friends are nerds, but I'm not.
www.maxmilne.com
Reposted by Max Milne
arxiv.org
September 18, 2025 at 8:16 PM
and then read them back out loud to me, and I interject to make edits. And then we flip roles. It is a drama and my mind loves it. Super interested to see where it all goes with us finding our individual flow state enablers. First trick? Having enough self awareness to know what gets the juice going
September 14, 2025 at 3:18 AM
Hello Brian, right there with you. This has been more approach for a couple of years. My thought has been this: "How to create a system that allows me to stay in flow states when working?". My best example in the real world is when I have a conversation with someone who 'gets me' and they take notes
September 14, 2025 at 3:16 AM
”Is that a film camera?”
September 7, 2025 at 10:18 AM
September 7, 2025 at 10:14 AM
According to wiki this has been being built for decades. en.m.wikipedia.org/wiki/Ch%C5%A...
Chūō Shinkansen - Wikipedia
en.m.wikipedia.org
September 3, 2025 at 12:12 PM
Only downfall?

It, as yet, doesn't have saved local projects. Once you quit VSCode that chat disappears. Is a bummer. And if introduced would be awesome.
September 2, 2025 at 3:31 AM
1. Easy undo button after each action taken, for if some code doesn't work. Awesome.
2. Instruction following is spot on. Hasn't 'taken its own lead' as Claude has a habit of doing.
3. Produces cleaner code.
4. Can be used on mobile via the ChatGPT app with cloud based work flows.
September 2, 2025 at 3:30 AM
Hello there, so your app can build multiple agents that connect together in a work flow? Or are they all individual agents that do one particular thing? Cheers,
August 8, 2025 at 2:20 PM
I imagine fear is the driver underneath it all for folks. An honour it both ways.
August 8, 2025 at 8:33 AM
Others have shared only relates to the ChatGPT interface. API, you can connect directly.
August 7, 2025 at 8:56 PM
I'm getting about 40 tokens on a 32GB M1 Max in LM Studio
openai/gpt-oss-20b

MXFP4
12.11 GB
August 6, 2025 at 10:48 AM
Reads very similar to @emollick.bsky.social tutoring prompt .
July 29, 2025 at 11:29 PM
Hello Ethan,

How are you accessing said model? Is it on OpenRouter as an 'unamed' LLM?
July 28, 2025 at 7:22 AM
Reposted by Max Milne
I think that's another point in favor of the analogy, personally: industry trying to replace programmers with LLMs is stupid in the same way that trying to replace carpenters with table saws is stupid
July 6, 2025 at 3:19 PM
Or you can just use this legends Python Code - github.com/fussbanana/o...
GitHub - fussbanana/obsidian-frontmatter-tool
Contribute to fussbanana/obsidian-frontmatter-tool development by creating an account on GitHub.
github.com
June 20, 2025 at 2:33 AM
I use Roo-Code and it has a Context Compression tool that is essentially a prompt to compress all context down. I find that Claude Code has this built in and compresses automatically. Try it out?
June 19, 2025 at 5:48 AM
These are two of the cutest sentences back to back.
June 19, 2025 at 5:46 AM
Glad to see you are back the blueSky train.
June 18, 2025 at 7:04 AM
Yep. Going through and making the changes. How do you do it swiftly?
June 6, 2025 at 3:16 AM