notes
Quick thoughts and TILs. Less polished than blog posts.
notes
Project Hail Mary: No Tragedy Tax@Observable YouI Wrote 40 Rules for My AI. They Cost More Than My Bugs.I Didn't Know My AI Could Run Without MeFixing Claude Code OAuth token revoked when switching accountsYour saved places say more about you than your dating profileShared instruction files are the env vars of AI agentsThe 40% Rule: What Running 4 Agents in Parallel Taught Me About Token EfficiencyVouch for agents: trust scoring for non-human contributorsStop generating, start composingThe hard part isn't harder — it's visible nowMy workspace is an operating systemInbound vs outbound: the two kinds of agents you need for open sourceScouting GitHub's agentic workflows: what agents forget about their own logsWhen Your AI Builds Its Own Agent FrameworkCross-pollinating open source: what a trust system can learn from an issue trackerThe Sherlocking Linepolars scan_parquet is 8-11x faster than read_parquet for filtered queriesyour agent isn't hallucinating. your specs are dead.claude code ships auto-memory. here's how to make it smart.the future of agentic coding is bottlenecked by a tmux split-pane race conditionmemory-sync: the missing feedback loop for Claude Code memory