Author
Dani Reyes writes DevMoment field notes on shipping real software with AI coding agents. Present tense, one repo at a time.
Quick answer (2026): I ran the same feature through GitHub Spec Kit, Kiro, and a plain plan.md in Claude Code to see which spec-driven setup actually earns its place. Spec Kit wins for writing shared project rules, Kiro wins for killing spec-and-code drift, and a hand-written plan.md wins on setup cost. The plain plan.md is the one I kept.
I hit Claude Code's weekly limit mid-refactor, then ran the same repo through seven alternatives for two weeks. The three I kept, and the honest catch nobody markets.
A month of running Cursor and Claude Code on the same repo. The honest 2026 verdict: it is not either-or, and the real question is not IDE vs CLI.
Claude Code checkpoints and /rewind are fast in-session recovery, not version control. A field log on what a checkpoint actually captures, what it silently ignores (bash writes, your own hand-edits), and the two-line git habit that keeps it safe.
The three arrays, the precedence rule nobody explains, the two gotchas that cost me an afternoon, and the exact settings.json allow, ask, and deny block I run now.
An honest 2026 field log on Claude Code skills: I installed 30 community skills, deleted almost all of them, and kept the three I wrote for my own repo. What a skill is actually good for, and the rule I use to pick skill vs CLAUDE.md vs MCP.
I lived on Claude Code Router for two weeks in 2026. Here is the per-task routing config I actually run, and where routing to cheap models quietly wrecks a session.
AGENTS.md vs CLAUDE.md is not a format war. A 2026 field log on running both files in one repo, why Claude Code ignores AGENTS.md, and the one-line import that fixed it.
A working reference: the five MCP servers I keep loaded in Claude Code after a year of daily use, the ones I removed, why every server costs context-window tokens, and the exact scope config that keeps the list short.
A field note on Claude Code plan mode after months of daily use: when it saves me, when I skip it, three failure modes the listicles miss, and the exact habit I built around it.
Six AI coding agents sit in my dock in 2026, but I do not open all six every day. Here is the honest field log of which one I reach for when the task is a refactor, a chore, or a tight edit loop, plus the routing rule that keeps surviving.
Is vibe coding bad? After six weeks of vibe coding a side project, here is where it helped, where it quietly made me worse, and the four guardrails I run now.
