The Best Cursor AI Settings for Maximum Productivity (2025)

Show Notes

This video shares Parker Rex’s practicalCursor AI setup for maximum coding productivity, with concrete steps you can copy-til-you-finish a feature—from per-project rules to indexing, docs, and cleanup.

Quick setup overview

• Use Cursor with a solid update workflow (Homebrew recommended) to keep features and fixes current.
• Build project-specific AI rules and organized documentation indexing to match your workflow.
• Tidy up after a feature to keep Cursor snappy and accurate.

Keeping Cursor up to date

• Install/update method:
  • Use Homebrew to install or update Cursor.
  • If you don’t have Homebrew, you can install it via Warp and run the appropriate command to install Homebrew.
• Why this matters:
  • Change logs can miss incremental updates; Homebrew helps ensure you’re running the latest improvements.

Rules and prompts strategy

• Cursor Rules (pre-prompts) are critical to shape behavior:
  • Access Rules via the top-right settings (Everything AI).
  • Don’t rely solely on the built-in rule prompt; create a dedicated rules folder with versioned prompts.
  • Use a Cursor Directory (cursor.directory) for boilerplates and templates, then customize per project.
• Practical example:
  • Start with a TypeScript-oriented rule set and tweak to taste.
  • Maintain different rule variants for different workstreams saved as separate prompts in a folder.

Indexing and docs: keeping context accurate

• Codebase indexing
  • The index helps Cursor know your entire codebase during edits (e.g., large refactors, many files).
  • If the codebase changes a lot, delete the index and re-sync periodically (recommended every few days).
• Documentation indexing
  • Create a docs catalog for external APIs, frameworks, or services you integrate with (Google Calendar, Whoop, Garmin, etc.).
  • Use kebab-case for keys and prefixes to ensure reliable tagging and lookup.
  • For each API, split into two tracks when needed:
    • API reference docs (specific endpoints/methods)
    • API docs (overview and usage guides)
  • How to add a new doc:
    • Add new Doc, paste the route/URL, verify the prefix, submit.
    • See status: yellow while indexing, green when done.
• API-specific details
  • Break down complex APIs into resource-based sections (e.g., accessControlList, calendars, channels) so Cursor can index and reference them accurately.
  • Example: for Google Calendar, create a dedicated resource map and index its pages.

Special cases for multi-resource integrations

• Supabase (and similar multi-product APIs)
  • Create separate docs for distinct SDKs or real-time features, e.g.:
    • supabase-js-sdk
    • supabase-real-time
• External docs vs API docs
  • If you have both, treat them as separate docs so Cursor can properly route queries to the right source.

Web search vs. offline, explicit prompts

• Default to not searching the web automatically:
  • Always turn off web search by default to avoid noisy or outdated results.
  • When you need fresh data, supply an explicit prompt with the exact outcome you want.
  • Include relevant docs or URLs in your prompt to anchor the AI.
• Prompt structure tips:
  • Provide a numbered user journey or task outline.
  • List the exact docs you’ve indexed or plan to reference, with prefixes where applicable.

UI/UX and performance tweaks

• Visual and interaction tweaks:
  • Turn off the fade chat stream for readability.
  • Disable the narrow scroll bar if it slows down reading long code blocks.
  • Turn off auto-scroll chat; read the generated content and verify before accepting.
• Model and feature toggles (examples from the video):
  • Use a targeted model setup for TypeScript projects (e.g., “Sunet” with “Elite” + React).
  • Prefer higher-quality code-models; selectively disable less relevant models.
  • Try specific presets like “cloud 3” and “O1 mini / 01 preview” if available, but be prepared to revert if stability is lacking.
• Performance housekeeping:
  • After completing a feature, delete related chats and composers to keep the session clean.
  • If Cursor slows down, clear caches:
    • Example command (macOS): rm -rf /path/to/Cursor/cache
    • Clear index or other temp data as needed.

Maintenance habits for long-term productivity

• Clean up after finishing a major feature to prevent bloat.
• Periodically refresh indexes and docs to keep references accurate.
• Maintain a small set of robust, per-project rule templates to minimize setup time for new workstreams.

Actionable takeaways

• Set up per-project rule directories early and keep them versioned.
• Regularly refresh codebase and API docs indexes (every few days).
• Index external API docs with kebab-case names and separate API docs from reference docs.
• Turn off automatic web search; use precise prompts with explicit docs references.
• Keep UI snappy: disable nonessential UI features and routinely clear cache and chats after big iterations.

Links

• Cursor Directory - Cursor rules and MCP servers
• Homebrew - Package manager for macOS
• Warp - Intelligent terminal with AI
• Google Calendar API - Calendar integration docs
• Supabase Docs - Backend-as-a-service documentation
• Cursor Docs - Official Cursor documentation
• Google Calendar API Reference - API resource documentation

If you want a deeper dive into any single section (e.g., building a robust per-project ruleset, or a step-by-step for Google Calendar API indexing), tell me which area you want expanded.