The public workbench of a one-person AI studio: agent harness engineering, vibe-coding practice, and study notes.https://octoblog.pages.dev/en-ushttps://octoblog.pages.dev/en/posts/consumption-agent/https://octoblog.pages.dev/en/posts/consumption-agent/Every AI agent you've heard of is a production agent. The complementary category — an agent that represents your attention, not your wallet — barely exists and, until now, didn't have a name. Here's the criterion, the eighty-year lineage, the loyalty problem, and what I'm building.Thu, 09 Jul 2026 00:00:00 GMThttps://octoblog.pages.dev/en/posts/context-compact-ate-my-constraints/https://octoblog.pages.dev/en/posts/context-compact-ate-my-constraints/I told the agent "don't touch module X" at the start of the task. Two hours in, it modified X — the constraint had been eaten by context compression. The mechanism-level root cause, and three fixes you can copy: externalized specs, discriminating compression, post-compact self-checks.Thu, 09 Jul 2026 00:00:00 GMThttps://octoblog.pages.dev/en/posts/one-person-studio/https://octoblog.pages.dev/en/posts/one-person-studio/In six months I shipped 50+ repos solo: a desktop workbench, a Mac App Store app, enterprise platforms, a dozen agent skills. This isn't a hustle story — it's the operating method: deterministic orchestration, never trusting self-reports, graded failure recovery.Wed, 08 Jul 2026 00:00:00 GMThttps://octoblog.pages.dev/en/posts/model-harness-agent/https://octoblog.pages.dev/en/posts/model-harness-agent/Put the same model into Claude Code and into a bare ReAct loop and usability differs by an order of magnitude — the entire difference is harness. Here are the six jobs a harness does, and why it's the real battleground of agent products.Mon, 06 Jul 2026 00:00:00 GMT