v2.2 is out. Some of the best parts aren't mine.
This release came out of the issues, the bug reports, and in a few cases other people's code.
The main change: you decide how your brain remembers now, not us. Seven controls in a settings window, in plain English rather than variable names. How much recent memories outrank older ones. What counts as a duplicate. How far to follow connections. Three options each, with an honest note about the tradeoff. Changes apply to your next search. No redeploy.
You can also change how memories get read for search. Pick a finer reading and it rebuilds what you have already saved. Resumable if you hit your daily AI limit, reversible until the last step.
Lose your password and the unlock screen is no longer a dead end. Sign in to Cloudflare and set a new one.
Credit where it belongs. Vincenzo Fabiano wrote the native menus, the Italian translation, and the About panel, and did the refactor that made the dashboard modular. The fix for Claude Code wrongly reporting Second Brain as unavailable is also Vincenzo's, off a report from maxacarvalho.
Desktop app v1.2 ships alongside: https://www.thesecondbrain.dev/download
One thing I am still unsure about. Every one of those seven controls has a sensible default and most people will never touch one. Useful, or complexity with better copy?
Replies
my instinct is complexity with better copy, honestly. i run a similar always-on memory setup for agent work and i've never once opened a settings panel after the first week - whatever the default recency/dedup behavior was on day one is what it still is for me. the seven controls read like they're really for the 5% of power users who hit a specific bad case (duplicate spam, stale connections surfacing over recent ones), not something the median user tunes proactively. i'd bet you get more mileage long term from picking better defaults per use case (coding vs research vs journaling) than from exposing the knobs, even with good copy.
useful, but i would not defend the settings window as the place for them.
seven controls with three options each is over two thousand configurations, and the person who tells you search got worse cannot tell you which one they are in. they changed something six weeks ago, or they changed nothing and their data changed. either way the report arrives without the setting attached and you are debugging a brain you cannot reproduce.
gal's 5% is right, and the awkward part is that those people do not know they are in it. so the moment a control is worth offering is not when someone opens settings, it is when a search comes back thin, in context. this brain has a lot of near duplicates, try loosening.
not an argument for fewer of them. i would rather have seven than one opinionated default that is wrong for my data.
when someone tells you a search was bad, can you see which of the seven they were on at the time?