Turn Repeated Workflows into Custom Slash Commands: Arguments, Live Bash Context, and File References
Take a small working chore-rotation app and package the three requests you keep retyping as custom slash commands in .claude/commands/: a /test-report whose ! bash line injects the live unittest run into context under a scoped allowed-tools grant, an /add-member that feeds a name through $ARGUMENTS into a precise data edit, and a /release-check that pulls a checklist in with an @file reference and walks it against the repo. A local self-check parses your real command files, executes your ! line in a sandboxed copy of the repo, and writes your proof-of-work evidence only when the commands genuinely work.
1 hr
Est. time
4
Outcomes
5
Rubric criteria
65%
Pass score
What you'll learn
Skills you'll have real reps in after shipping this.
The scenario
You share a small repo (ChoreWheel, a household chore rotation app) and every Claude Code session in it starts with the same three requests typed out by hand: run the tests and tell me what broke, add this new member to the seed data the right way, walk the release checklist before we tag. Retyping a paragraph of instructions is slow, and each retype drifts a little from the last, so the fourth time you ask for the checklist walk it checks different things than the first.
Custom slash commands fix that. A command is a Markdown file in .claude/commands/ whose body becomes the prompt the moment you type /its-name, and this task makes you use the whole command toolkit on a real repo: frontmatter metadata (description, argument-hint, allowed-tools), argument substitution with $ARGUMENTS, live bash context injection with a ! line, and @file references. You do not just write the files: the kit's self-check executes your actual ! line in a sandboxed copy of the repo and only writes your evidence when the suite really runs, and you paste one invocation from your own live session as proof you drove the commands for real.
Your role
You are the engineer who owns the Claude Code setup for a shared repo. Your deliverable is three working slash commands that package the repo's repeated workflows, plus a session transcript and a short note drawing the line between commands and skills, packaged as a single submission.
Start the task to unlock the full brief
You'll get the step-by-step requirements, setup commands, the 5-criterion grading rubric, tips, and the ability to submit your solution for instant AI grading.
Free to start · submit when you're ready
Build custom Claude Code slash commands on a real repo
Custom slash commands turn the instructions you keep retyping into named, shareable prompt files, and this task makes you build three that a real repo needs. You work inside a small working chore rotation app and author /test-report (a ! bash line injects the live unittest run into context under a scoped allowed-tools grant), /add-member (a $ARGUMENTS substitution feeds the new name into a precise data edit), and /release-check (an @file reference pulls the checklist in for a walk against repo state). A local self-check parses your actual command files and executes your bash line in a sandboxed copy of the repo, so what you submit is machine-generated proof the commands work plus a transcript from your own session. You come away fluent in the full command toolkit (frontmatter, arguments, bash context injection, file references) and clear on where the command vs skill boundary sits.