MirrorEthic

4 ms·

1. 1.▲
   Show HN: Cl33-opLM, a 236M language model with a reversible operator bottleneck(cl33.t3atlas.dev)

   3 points by MirrorEthic 2mo ago|0 comments
2. 2.▲
   by MirrorEthic 9mo ago

   Sorry about the late reply!
3. 3.▲
   by MirrorEthic 9mo ago

   Good question. I don't have a formal benchmark yet, but here's what I've measured in practice: Token reduction: 64-95% depending on codebase size and work pattern. The variance is because it depends on how many files are in y
4. 4.▲
   by MirrorEthic 9mo ago

   v1.1.0 Just Dropped 33k views and 80+ stars in 24 hours — thank you. Your feedback drove this: What's New Attention History Tracking Every turn now logs to `~/.claude/attention_history.jsonl`. Query your trajectory: ```bash p
5. 5.▲
   by MirrorEthic 9mo ago

   CLAUDE.md is static same content every session (and a soft 40k character limit) This is dynamic attention routing. Files get scored based on what you're actively discussing. mention "auth" and auth-related docs go HOT (full i
6. 6.▲
   Show HN: Claude Cognitive – Working memory for Claude Code(github.com)

   15 points by MirrorEthic 9mo ago|7 comments
7. 7.▲
   by MirrorEthic 9mo ago

   I built Claude Cognitive because Claude Code kept forgetting my codebase between sessions. The problem: Claude Code is stateless. Every new instance rediscovers your architecture from scratch, hallucinates integrations that don't exist