LevelUpWith

Explore / Software & Engineering

Cursor for Real Software Work: From VS Code to Agents

Illustrated cover for the course “Cursor for Real Software Work: From VS Code to Agents”

Get from VS Code user to confident Cursor practitioner. You will use tab completion, inline edits, chat and the agent on a practice repository, steer them with rules, MCP and codebase indexing, run background agents and Bugbot reviews, and make informed choices on models, cost, privacy and team setu

Capable · 18 levels · 2 free · Created Oct 2026 · Professionally curated by levelupwith.com

What's inside

  1. Level 1: Moving In: Install Cursor and Import VS CodeFree
    Install Cursor, import your VS Code extensions, settings and keybindings, and clone the practice repository. The course is independent of Anysphere, the company that makes Cursor, and this level shows you how to check its claims against Cursor's current docs and changelog.
  2. Level 2: Tab Completion That Earns Its KeepFree
    Learn how Cursor's Tab completion suggests multi-line edits and cursor jumps, and when to accept part of a suggestion, all of it, or none.
  3. Level 3: Inline Edits on a Selection
    Use the inline edit prompt to rewrite a selected block of code or generate new code in place, then read the diff before you keep anything.
  4. Level 4: Chat with Context: @-Mentions and Ask Mode
    Use chat in Ask mode to ask questions about the code without changing it, and steer its answers by @-mentioning the files, folders, docs and symbols that matter.
  5. Level 5: Codebase Indexing and .cursorignore
    Learn how Cursor indexes your codebase for semantic search, how to check the index is current, and how to use .cursorignore to keep files out of indexing and AI context.
  6. Level 6: Meet the Agent: Modes and Your First Task
    Learn how the agent plans and makes multi-file changes, how Agent, Ask and Plan modes differ, and how to frame a task so the agent stays in scope.
  7. Level 7: Reviewing Agent Diffs: Accept, Edit or Reject
    Build a review routine for agent changes: read the multi-file diff, use checkpoints to roll back, and decide hunk by hunk what to keep.
  8. Level 8: Terminal Commands and Safe Approvals
    Control which shell commands the agent may run, using approval prompts, allow and deny lists and sandbox settings, so it can work quickly without doing anything destructive.
  9. Level 9: Rules Files and Project Instructions
    Steer the agent with project rules in .cursor/rules and AGENTS.md, choosing when each rule applies, and check that the rules actually change its behaviour.
  10. Level 10: Connecting MCP Servers
    Give the agent extra tools and data sources by connecting Model Context Protocol (MCP) servers, and weigh what you gain against what each server can access.
  11. Level 11: Tests and Debugging with the Agent
    Use the agent in a test-first loop: reproduce a bug with a failing test, have the agent fix it until the tests pass, and check it didn't cheat.
  12. Level 12: Choosing Models and Understanding Your Bill
    Pick models for each task (Auto, fast models, frontier models, Max Mode) and work out from Cursor's current pricing page what usage-based billing does to your monthly cost.
  13. Level 13: Privacy Mode and Data Handling
    Configure privacy mode and understand, from Cursor's current security and privacy documentation only, what happens to your code, prompts and index data.
  14. Level 14: Cloud Agents for Long-Running Work
    Hand long tasks to Cursor's cloud agents (formerly called background agents), which run in remote environments, and manage them in parallel from the Agents Window.
  15. Level 15: Bugbot Code Review on Pull Requests
    Add Bugbot automated reviews to the practice repo's pull requests, tune it with review rules, and sort its findings into fix, dispute or ignore.
  16. Level 16: Team Settings and Shared Practice
    Set up Cursor for a team: admin and privacy controls, shared rules and MCP config in the repo, usage limits, and a written policy for AI-assisted changes.
  17. Level 17: Cursor vs Claude Code, Codex and Copilot
    Compare Cursor with Claude Code, OpenAI Codex and GitHub Copilot on workflow, context handling, agents, review, pricing model and privacy, using each tool's current documentation, so you can choose the right tool for a job.
  18. Level 18: Final Challenge: Ship a Feature End to End Exam
    Use every skill from the course to deliver a feature in the practice repository, from context and rules through agents, tests, review, cost, privacy and team configuration.

Access

The first 2 levels are free with a free account. Every level, the podcast edition and the AI tutor come with All Access at £4.99/month or any Creator plan — see pricing.

Start free →