@trq212 I have an identity file that gets injected to Claudeβs context with session start hook. The same file also contains an instruction at the end to save new info about me to it when I share something new any session.
The only problem I have right now is the 10k character limit. π₯²
A Post Specific to @claudeai Models:
I used to worry about which model to use for a specific task, but then I realised that humans should always be facing the most capable model. In my case, that is Fable, the most capable one at understanding INTENT.
Why? I have seen people complain about Opus or Sonnet not working as expected. But the thing is, as we go down the rungs, the prompt quality starts mattering. Try this: Ask Fable to delegate and see how well those same models perform with Fable's instructions. You will be amazed.
And that is how I am protecting the expensive AI model's quota: by letting it delegate, not by replacing it. The top model plans and writes detailed instructions, a cheaper model executes them, and the top model verifies the result against the actual files. Like a good engineering team.
AI models are inherently probabilistic, so if you asking your AI to type front-matter by hand in your markdown notes, you might run into trouble. Even if you have templates for every note kind, the problem doesn't disappear.
Each AI model (even different instances of same models) formats YAML a little differently, and that drift compounds over hundreds of notes, which I had to learn the hard way.
A code can apply templates deterministically; the AI only decides what goes in the fields. This is exactly what I have adopted.
If you are using a plain markdown folder as your second brain and still want to use something like Things 3 to manage your tasks, you can give AI full control of Things 3 and have AI fill in tasks for you.
Cultured Code themselves say AppleScript is one of the safe paths. But AppleScript dictionary has holes: checklists, headings, and repeat rules are simply not in it, so the AI cannot see or touch them. Things' own .sdef file inside the app bundle doesn't have the commands. Before trusting an AI integration, know what your AI cannot see.
@karpathy I just added a line in my CLAUDE.md that I dictate using Wispr Flow, and to look for intent not just literal meaning. When Claude reports back what it understood, it almost never gets in wrong. π«
Tried the new terminal mode in Kite @zerodha and @AngelOne Scalper Mode. It is really nice that scalpers are finally getting recognition and essential features.
I just hope @zerodha also introduces βORDER PRESETSβ like AngelOne has done.
And @AngelOne please fix the %SL preset.
Traders and investors have their own preferences on what they want on their screens. An intraday scalper will value speed and a quick order window above all else, while an options trader may always want the option chain open.
Kite's classic web UI was designed to work universally for everybody. More than a decade later, despite us adding a ton of features and capabilities, Kite is still extremely light and fast with a tiny bundle size.
While desktop trading platforms have historically had widgets, web technologies weren't suited for this for the longest time. We've prototyped this on and off over the years, and last year building a web-based trading terminal became doable with native web technologies.
Introducing Kite Terminal mode. We rebuilt Kite web completely and made every single feature it has into a widget. You can now drag, drop, and rearrange all the components inside Kite to design your own interface tailor-made for the way you trade and invest. We have put in a ton of work to make the interface snappy with smooth UX. Now you can build your own trading interface like assembling Lego blocks.
In addition to investment and trading widgets, we've also added a bunch of productivity and personalization ones like notes, clocks, calendars, wallpapers, my Twitter feed (follow me on Kite) π and more.
I'm excited to see how people design their layouts. Share a screenshot of your workspace in the comments.
@Shpigford Obsidian might be good for you if you decide to add a few notes here and there.
But your requirements seem pretty much solvable with just Google Drive. They have all three of those. I use Drive search inside my Obsidian vault too, basically for PDF and image OCR.
@vanschneider@mymind I have seen version history for notes in many apps. But why hasnβt anyone thought of building it into the time meta data itself π€―
Hats off π«‘
@TfTHacker@kepano He is mostly focusing on the fancier things inside @obsdmd and entirely misses the point that the core of the system (the notes, pages in the notebook) stay intact and readable even if the entire ecosystem burns down.
@itsolelehmann Why use IDE? Why softwares focus on UI? Why use emojis? Why use display panels on machines? Why watching movies is easier than reading the same story?
You can skip all of them along with @obsdmd, but itβs just that the human brain has evolved to process visuals much much better.
This is an exercise from the last chapter of my next book "The Law of The Sublime". Try it out.
This clip is from my conversation with @sariazout@wwwsublimeapp
@phntm_pie @fortelabs Totally agreed! I felt the same since I first watched this video.
The thing is daily habits are low stakes. You do them or miss them, still catch up on next day. But just once a year increases the stakes too much with a lot of work.
Listening to Tiny Experiments by @annelaure and realised something small but powerful about @NickMilo's ACE framework.
Calling things "efforts" instead of "projects" quietly changes how you think about them. Projects need endpoints and milestones. Efforts just need... effort.
Suddenly an experiment like "I will do X for 30 days and see what happens" fits naturally. It's not a project as there's no deliverable. But it's clearly an effort.
One word. Whole new space opened up.