Cursor Rules & Prompts

This project uses Cursor rules and skills so the AI follows architecture, workflow, and quality checks. The full reference lives in the repo at docs/CURSOR-RULES-AND-PROMPTS.md; here is a short overview.

Always-on workflow

For any code change (fix, feature, refactor, review, debug), the agent:

1. Triggers the right rule from agent-triggers (subagents, code-review, debugger, idea-to-feature).
2. Plans first: scope, approach (architecture, layers, standards), order of steps.
3. Implements per the plan and triggered rule.
4. Runs npx tsc --noEmit and fixes errors.
5. Audits changed code (correctness, layers, security, types).
6. Finalizes with summary and how to verify.

For documentation: when adding or changing features, the agent also updates Docusaurus docs (see rule docs-sync.mdc).

Subagents (one rule: subagents.mdc)

Intent	Section	Skill
Refactor folder structure / migrate to app-core-features	Refactor	refactor-to-architecture
Add new feature / module / screen flow	Add Feature	add-new-feature
Architecture or structure review	Architecture Review	architecture-review

Skills live under .cursor/skills/. The agent picks the section by intent, runs the skill, then runs npx tsc --noEmit (except for review-only).

Other rules

• code-review-audit.mdc — Code review / audit; checklist and report format.
• debugger.mdc — Reproduce → isolate → fix → verify (tsc + user confirmation).
• idea-to-feature.mdc — New idea from Figma/image/URL: capture → plan with codebase → suggest enhancements → implement.

Quick prompts

• Refactor the project to the new folder structure.
• Add a new feature for [name].
• Do an architecture review of the project.
• Do a code review of the codebase.
• Debug this: [error or behavior].
• I want this feature. [Figma URL / image]. Plan it with our codebase, suggest enhancements, then implement.

Full table and details: docs/CURSOR-RULES-AND-PROMPTS.md in the repo.