Skills are slash commands that give your AI assistant (Claude Code or Cursor) structured workflows. Type them in the chat to trigger the guided process.
Skills location
Skills are stored in .claude/skills/. Each skill is a Markdown file with instructions the AI follows. You can read, fork, or extend them.
/new-toolTransforms this boilerplate into a fully functioning internal tool by walking you through a guided questionnaire. Generates a spec, waits for your approval, then scaffolds the entire application.
When to use
Use this once, at the start, to bootstrap your tool.
What it does
Example interaction
/new-tool"I'm going to help you create a new tool from this boilerplate.
I'll ask you some questions, then generate a spec for your review.
Let's start with the basics.
What is the name of your tool?"/add-featureScaffolds a new page, API route, and optional Prisma model into an existing tool — following the patterns already present in the codebase. Consistent with whatever /new-tool generated.
When to use
Use after /new-tool, whenever you want to add a new capability to your tool.
What it does
Example interaction
/add-feature"I'll help you add a new feature to this tool.
What is the name of this feature?
(e.g. 'Invoice overview', 'User management', 'Export report')"/health-checkAudits your tool for bugs, missing error handling, security issues, and boilerplate-specific problems. Produces a prioritised report of findings — does not apply fixes automatically.
When to use
Run before a deploy, after a sprint, or when something feels off.
What it does
Example interaction
/health-check"Running a health check on this tool.
──────────────────────
🔴 CRITICAL (1)
- app/api/export/route.ts: No session check — unauthenticated users can trigger export
🟡 HIGH (2)
- app/api/users/route.ts: Request body not validated with Zod
- lib/db/index.ts: Raw SQL query on line 42 — use Prisma instead
🟢 LOW (1)
- app/dashboard/page.tsx: No empty state for the data table"/improveReviews your tool for UX improvements, performance optimisations, and missing features — based on what the spec says the tool should do. Produces a prioritised suggestion list.
When to use
Use at the end of a sprint, before a demo, or when the tool works but could be better.
What it does
Example interaction
/improve"I'll review this tool for improvements.
Reading the codebase now.
──────────────────────
✨ Suggestions (4 total)
1. [UX · Quick] Add loading skeletons to the data table — currently shows nothing while fetching
2. [UX · Medium] The form has no client-side validation — users only see errors after submit
3. [Feature · Medium] The spec mentions CSV export but it's not implemented
4. [Performance · Hard] The dashboard fetches all records — add pagination or limit"A skill is just a Markdown file. Create a new folder in .claude/skills/my-skill/ with a SKILL.md inside. Structure it with a description, a questionnaire, and a set of instructions for the AI to follow. Type /my-skill in the chat to invoke it.
# My Custom Skill
This skill does X.
**When to use:** Describe when to trigger this.
---
## SECTION 1: INTRO
Say: "I'll help you do X. Let me ask a few questions."
## SECTION 2: QUESTIONS
Ask one at a time:
1. "What is the name of X?"
2. "What should X do?"
## SECTION 3: EXECUTE
Based on the answers, [do the thing].The following skills are part of the FRMWRK development plugin and form a workflow for planning and executing work. They're not built into the boilerplate but work seamlessly with it.
| Command | Purpose |
|---|---|
/check-status | Compact status overview of the project — tests, branches, recent changes |
/plan-next-step | Generate a technical plan for the next step, ask critical questions, wait for approval |
/give-feedback | Process feedback or questions during an active plan — clarify, update, re-approve |
/executing-plan | Execute an approved plan step-by-step with checkpoints and task tracking |
To use these skills: install the FRMWRK development plugin for Claude Code