Blog — notes on agent-native workflow and prompt-composed systems

2026-09-16

Kanban for AI agents, proven in daily dev work — kanbento v0.3

Kanban for AI agents in practice: an AI harness using kanbento to track and deliver real work across a dozen repos — with the numbers.

2026-07-24

Grilling is uncertainty reduction — why grill-me works, and how it compounds

The viral grill-me skill got called 'adding friction.' That's the mechanism misnamed. The relentless questions buy certainty — an entropy drop before commitment — and the practice only compounds if the answers outlive the session.

2026-07-20

Make policies explicit, or your agents will improvise them

Humans fill policy gaps with judgment. Agents fill them with confident improvisation. Kanban's 'make policies explicit' becomes load-bearing — as legible workspace artifacts, not tribal knowledge.

2026-07-16

The AI Productivity Paradox: Flow Metrics for Agentic Delivery

AI lifts individual output and leaves system delivery flat — the productivity paradox nobody can measure. Issue #1 of an ongoing report: our board's real flow numbers, definitions and all.

2026-07-08

Spec-driven development works — until the spec stops being true

Every SDD guide says the spec should be a living artifact. No tool in the wave actually delivers that — because 'living' is a process property, not a document format. That's what a board is for.

2026-06-29

Instruction Bleed: when your SKILL.md files quietly interfere

A new arXiv paper names a real failure mode: edit one prompt module and an unrelated one's behavior shifts. What it measured — and what it doesn't claim.

2026-06-29

What Instruction Bleed means if you write SKILL.md / AGENTS.md

The practitioner's read on instruction bleed: where cross-module interference is most likely to bite, and the cheap check the paper hands you.