Repair the workflow
with a slash of gold.
The Japanese art of kintsugi mends broken pottery with lacquer dusted in gold — refusing to hide the seam, and treating the break as part of the object's history. Today's Claude Code feature follows the same logic for your terminal.
Where a workflow keeps cracking, pour /gold on it.
Custom slash commands let you turn a repeatable prompt — a code review checklist,
a release-notes generator, a runbook step — into a one-keystroke ritual. Drop a
Markdown file into .claude/commands/ and Claude Code surfaces it as
/your-command in the prompt picker, scoped to the project (or to your
whole user with ~/.claude/commands/).
The file is the prompt. Frontmatter pins the model and the tools it's allowed to
touch; $ARGUMENTS threads in whatever you type after the slash.
Crack-prone process, golden seam.
# .claude/commands/postmortem.md --- description: Draft a blameless postmortem from the last incident allowed-tools: Read, Grep, Glob, Bash(git log:*) model: claude-sonnet-4-6 --- Walk the repo for changes since $ARGUMENTS, pull the matching alert window from runbooks/, and produce a postmortem in our standard template — timeline, contributing factors, customer impact, action items. Be specific. No blame.
Then, in your session: /postmortem 2026-05-09T14:00Z. The seam shows —
and that's the point. Anyone on the team can read the command, tweak it, commit it.
The repair is part of the repo's history.
Treat the break not as the end of the bowl, but as the place the gold goes. — a working translation