mvyshnyvetska

4 ms·

1.
Show HN: Breathe-Memory – Associative memory injection for LLMs (not RAG)(github.com)
6 points by mvyshnyvetska 7mo ago|1 comments
2.
by mvyshnyvetska 9mo ago
Yeah, that's more process-tooling territory — keeping this dev-focused for now.
3.
by mvyshnyvetska 9mo ago
Update: just shipped --brief flag. Thanks for the push!
4.
by mvyshnyvetska 9mo ago
Fair points :). The naming is 'evolutionary' — started as 'semantic diff' because it analyzes meaning not just lines, but 'commit review' is more accurate for what it does now. You're right about the redun
5.
Show HN: Semantic-diff – understanding intent, risk and impact behind Git diffs(github.com)
2 points by mvyshnyvetska 9mo ago|6 comments
6.
by mvyshnyvetska 10mo ago
Interesting approach. A few questions: How do you handle false positives from mutation strategies like Base64 or token smuggling? In my experience, a lot of "successful" jailbreaks from automated fuzzing don't actually produc
7.
A three-layer memory architecture for long-running agents
1 points by mvyshnyvetska 11mo ago|0 comments
8.
by mvyshnyvetska 11mo ago
Hit exactly this problem some time ago (preliminarily buried 5 or 6 versions of memory systems). Generic summaries or RAG results often feel useless because models optimize for "what would be useful to explain to anyone" rather th
9.
by mvyshnyvetska 11mo ago
We've been solving similar problems differently. Instead of a flat progress.txt, we use three layers: semantic context (what are we building and why), structured intention tracking (blockers, dependencies, decision paths), and git for