最新

This is my attempt in building a memory that evolves and persist for claude code.<p>My approach is inspired from Zettelkasten method, memories are atomic, connected and dynamic. Existing memories can evolve based on newer memories. In the background it uses LLM to handle linking and evolution.<p>I have only used it with claude code so far, it works well with me but still early stage, so rough edges likely. I&#x27;m planning to extend it to other coding agents as I use several different agents during development.<p>Looking for feedbacks!

Looking for recommendations for sites similar to Downdetector

I’ve been exploring the risks of letting autonomous AI agents generate or execute code — especially via WASM.<p>Night Core is a lightweight console that treats execution as a boundary, not just output. It uses signature-based validation, human approval queues, and sandboxed runtimes (like Wasmtime) to let you review or reject what runs — before it happens.<p>Demo screenshots: <a href="https:&#x2F;&#x2F;x.com&#x2F;xnfinitecore&#x2F;status&#x2F;2010067157774991457?s=46" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;xnfinitecore&#x2F;status&#x2F;2010067157774991457?s=46</a><p>Still early, but open to ideas, pushback, and contributions. Curious if others are thinking about this edge.

I&#x27;m researching demand for an end-to-end encrypted social platform focused on private groups (family&#x2F;friends) rather than public feeds.<p>Core thesis: &quot;They don&#x27;t own you&quot; is becoming a movement. People want: - E2E encryption (platform can&#x27;t read content) - True deletion (cryptographic, not just &quot;marked as deleted&quot;) - Subscription model (no ads, no data mining) - Open source (verifiable claims)<p>The gap: Signal is for messaging, Discord isn&#x27;t private, Facebook is surveillance capitalism. Nothing serves the &quot;private social groups&quot; use case with real privacy.<p>Before building, I need to validate if this is a real pain point or just something that sounds nice but nobody would actually switch for.<p>2-min anonymous survey: https:&#x2F;&#x2F;forms.gle&#x2F;bfZYPfxMUBCc1iACA<p>Honest feedback appreciated—especially if the answer is &quot;this wouldn&#x27;t work because X.&quot;

Now that Claude et al practically made building anything and greenfield projects in particular 100x faster start l, how do you pick side projects?<p>Do you pick more ambitious ones? Do you try crazy things that you wouldn&#x27;t have dared before? Do you try all sorts of different ones in parallel? Do you pursue projects that can make you money? If so, do you ever validate or jump straight in?