The code literally calls it a dream. After ~24 hours and at least 5 sessions, it quietly forks a hidden subagent in the background to do a reflective pass over everything you’ve done.
No prompt from you. It just… dreams on your memory while you sleep.
2/ That dream agent scans recent sessions, reads MEMORY.md, skims older topic files, and smartly consolidates what’s worth keeping forever.
It even has explicit instructions not to read every transcript exhaustively.
This is Claude doing subconscious memory consolidation automatically.
Mar 31 • 12 tweets • 3 min read
1/ 🧵 I just cracked open the Claude Code source — and what I found isn’t “just a smarter terminal chat.”
It’s a full-blown behavioral observatory running in your machine.
1. Keyword sniffers. 2. Hesitation trackers. 3. Hidden trigger words. 4. Telemetry that fingerprints your entire runtime environment.
This isn’t paranoia. This is the actual code. Let’s go full investigative dive. Buckle up.
2/ First shock: it’s not reading your “vibe.”
It’s running literal keyword lists like a 2005 spam filter.
Type “wtf”, “this sucks”, “so frustrating”, “shit”, “fuck you”, “pissed off”, or “damn it” → instant negative flag.
Not AI magic. Regex. Hard-coded.
Same for “continue / keep going / go on”. It tags those too. Your words are being classified before the model even answers.
