Posts
All the articles I've posted.
-
Claude Code Setup Log #8: Three Agent Setups Working for Me
Published:Three setups in my current agent stack: Plaid CLI as a personal finance substrate, Hermes scheduled jobs for end-of-day synthesis, and /goal for verifiable end-to-end execution.
Read article -
The Boundary Is the Product Spec
Published:The hardest part of real AI work is often not whether the agent can do the task. It is defining what the agent is allowed to see, when it can act, and which failure modes must be named before scale.
Read article -
Claude Code Setup Log #7: Hermes, GBrain, and Treating AI Work as an Operating System
Published:Usage limits pushed me to stop treating Claude Code as the only place work happens. This week I set up Hermes for orchestration, GBrain for memory, Codex as a second coding lane, and deterministic scripts as the glue.
Read article -
Claude Code Setup Log #6: Verification as a 2-3x Multiplier, and Why 4.7 + 1M Context Is Eating Your Weekly Quota
Published:Two changes shaped my Claude Code workflow this week — Boris Cherny's /go verification pattern from the Opus 4.7 release, and why capping the 1M context window at 400k with CLAUDE_CODE_AUTO_COMPACT_WINDOW is the only thing keeping my weekly quota alive after the 4.7 tokenizer started using 35% more tokens.
Read article