Beads Is Excellent Until Two People Need the Same Board
A field guide to where Beads' local-first, Dolt-backed model stops paying off, and what a hosted tracker costs you in exchange.
Read itThe Trail blog
Field notes from working with coding agents
What breaks when an agent runs your project for months: the context that does not survive the session, the loose ends nobody wrote down, and the record of who actually changed what. Written while building Trail.
What the EU AI Act Actually Asks You to Record About AI-Written Code
A six-field record for AI-written code, the tools that produce it automatically, and why the deadline you heard about already moved.
Issue Trackers Built for AI Coding Agents, Compared
Seven trackers built for AI coding agents, scored on where the data lives, whether it remembers, who gets credited, and who else can see it.
CLAUDE.md vs AGENTS.md vs GEMINI.md: One Project, Three Agents
A field-by-field comparison of the three instruction files, plus three strategies for keeping Claude Code, Codex CLI, and Gemini CLI reading the same rules.
Running One Coding Agent Across Six Client Repos Without Context Bleed
A working setup for freelancers and small studios running one AI agent over many client codebases without instructions crossing between them.
How to Stop Your Coding Agent From Making the Same Mistake Twice
A concrete test for which fixes are worth writing down, and where to put them so the next session actually reads them before it acts.
todo.md Works Until the Third Session
The four ways a todo.md file breaks once a Claude Code project runs past one session, and the three properties whatever replaces it actually needs.
Your CLAUDE.md Is Too Long and Claude Is Quietly Ignoring Half of It
The working ceiling for CLAUDE.md, a five-minute test for which rules still land, and where the rest of the file actually belongs.