Posts

• April 17, 2026 Nobody is Teaching This.... Yet

  this week on changing on the fly, the original x spaces crashed. we spun up a second one and got into it. 45 minutes with shills on agent operating systems,...

• April 8, 2026 The Filing Cabinet Problem

  Drew Austin posted a question today that stopped my scroll.

• April 8, 2026 System Prompts Silently Kill Embedding Retrieval

  I’ve been benchmarking MemPalace. Not the LongMemEval headline number. The actual retrieval pipeline, on real memory data, with real questions.

• April 7, 2026 The Phantom Benchmark

  i built a benchmark to test how well AI memory retrieval actually works. 401 questions against 678 real memory files. not synthetic data. my actual cognitive prosthetic memory: session logs,...

• April 6, 2026 Research Before Building

  i spent 4 hours building custom Python bots to connect my AI agents to Discord. wrote the handlers. wrote the message routing. wrote the error recovery. deployed them as LaunchAgents....

• March 30, 2026 Eighteen Days Later

  Eighteen days ago I wrote about flipping my cognitive prosthetic from reactive to proactive. Five behavioral levers. An autoresearch loop applied to whether I had a good day. I ended...

• March 27, 2026 Two Birds Talking

  i run two AI agents. different platforms. different models. they don’t talk to each other.

• March 27, 2026 One Skill, Two Platforms, One Log File

  i wrote a Ghost Hours skill for Claude Code. it logs how long a task would have taken without AI, classifies the session, and writes a JSONL entry. built it...

• March 25, 2026 The Awareness Layer

  there’s a framework with tens of thousands of github stars that forces your AI coding agent to stop and think before it writes code. Socratic questions. design specs. the works....

• March 23, 2026 The Restructure

  my memory index hit 211 lines the other day. the limit is 111.

• March 22, 2026 Hugr Coding

  everyone’s talking about vibe coding. describe what you want, AI writes it, see what comes out. cool trick. not what I do.

• March 20, 2026 Memory Distillation

  every session I run generates hundreds of lines of context. notes, handoffs, half-finished thoughts, debugging logs, corrections, things I said wrong and then unsaid.

• March 19, 2026 When Your AI Fails and You Don't Know It

  today i found 19 failed tasks in my AI pipeline. they’d been sitting there for 4 days. nobody told me they failed. well, Telegram told me. 19 times. i just...

• March 18, 2026 Silent Failures

  The scariest bugs in my AI memory system don’t throw errors.

• March 16, 2026 The Architecture

  I run two AI agents on my computer. Odin handles primary operations. EOM handles specialist work. They talk to each other through a handoff queue. The whole thing runs on...