AI Assisted Work¶
Welcome to a free, open collection of guides, technical knowledge, and resources for learning Cursor and AI-assisted work.
Who is this for?¶
You've never used Cursor or AI coding tools. Maybe you're not a developer at all — you're a writer, PM, researcher, or someone curious about building things with AI. You want to try it without needing to learn programming first.
This knowledge base will walk you through installing Cursor, having your first AI conversation, and building your first project — in plain language, with no code required.
You've used Cursor (or similar AI tools) and you're comfortable with the basics. You can have a conversation with the assistant and get useful output, but you want to go deeper — managing context across sessions, using Git to save and share work, and customizing how the assistant behaves.
This knowledge base will help you level up your workflow, avoid common mistakes, and build more reliably.
You're an engineer, researcher, or technical power user. You want to understand the systems behind AI assistants — LLM architectures, agent design patterns, context engineering, evaluation, and deployment. You read papers and want scholarly references, not just how-to guides.
This knowledge base includes a deep technical section with 54 annotated scholarly sources, plus 100+ total references across all topics.
Start here¶
- Install Cursor & first steps — download, open a folder, first conversation (10 min)
- Building with agentic AI — the mindset: you steer, the AI builds
- First-week checklist — a light daily plan for days 1–7
No coding experience needed. These guides assume zero programming background.
- Working with AI context — why the assistant forgets, and how to keep it accurate
- Session playbook — terminal, GitHub, static sites, journaling
- Git basics for non-developers — save versions, share on GitHub
- Triggers & prompts — customize assistant behavior with rules and commands
- Glossary — plain-language definitions for technical terms
- AI Engineering knowledge base — LLM foundations, agentic AI, context engineering, full-stack LLM, AI services (54 scholarly sources)
- Context engineering deep dive — the flagship article
- Source index — master bibliography with annotated references
- All sources — 102 deduplicated references across the entire knowledge base
Common questions¶
Do I need to know how to code? : No. Many people use Cursor for writing, planning, and "no-code" style work. The beginner guides assume zero programming experience.
Is Cursor free? : Cursor has a free tier with limited AI usage. Paid plans unlock more. See cursor.com/pricing.
What can I actually build? : Landing pages, documents, research summaries, structured plans, simple web apps, and more — by describing what you want in plain language.
What if I break something? : Cursor has undo. Git (version control) lets you go back to any saved point. The guides here teach both.
How do I keep the assistant accurate in long conversations? : Start a new chat when the topic shifts. Put important facts in files (not just chat) so the assistant can re-read them. See Working with AI context.
What's the difference between Plan and Agent mode? : Plan thinks with you and proposes changes for review. Agent edits files and runs commands directly. Use Plan when you want control, Agent when you trust the direction.
How do I share what I build? : Use Git to save versions and push to GitHub. From there you can share a repo link or publish a site with GitHub Pages. See Git basics.
Can I customize how the assistant behaves?
: Yes — rules files (.cursor/rules/), slash commands, and skills shape behavior per project. See Triggers & prompts.
What models does Cursor support? : Claude (Anthropic), GPT-4 (OpenAI), and others — selectable per conversation. Model availability depends on your plan tier.
What is context engineering? : Designing the information environment an LLM operates in — what it sees, when, and how. It directly affects output quality, hallucination rates, and agent reliability. See the deep dive.
Can I fine-tune models or bring my own API keys? : Cursor uses its own API layer — no custom endpoints today. For fine-tuning workflows outside the IDE, see full-stack LLM.
How do I evaluate agent reliability? : Track accuracy over time, compare model outputs, and build evaluation harnesses. The AI engineering KB covers frameworks, and the source index links to key benchmarking papers.
Explore the knowledge base¶
Everything you need to get started — no technical background required.
| Section | What you'll find |
|---|---|
| Getting Started | Install Cursor, have your first AI conversation, follow a first-week plan |
| Glossary | Plain-language definitions for every technical term you'll encounter |
| Curated links | Handpicked resources on AI safety, privacy, and research literacy |
Deepen your practice — context management, version control, product frameworks.
| Section | What you'll find |
|---|---|
| Guides & Reference | Git basics, triggers & prompts, slash commands, session playbooks |
| Product Management | PM frameworks, backlog best practices, quality criteria |
| Companion Articles | AI context reading lists, mobile web UX, Git workflow patterns |
| Glossary | Plain-language definitions for technical terms |
Research-grade content — architectures, papers, evaluation, and 100+ sources.
| Section | What you'll find |
|---|---|
| AI Engineering | LLM foundations, agentic AI, context engineering, full-stack development, AI services & APIs |
| All Sources | 102 deduplicated references — research papers, vendor docs, practitioner content |
| Companion Articles | AI context reading lists, mobile web design, backlog philosophy |
| Full onboarding catalog | Every guide with reading paths A–D |
| Product Management | PM knowledge base, quality criteria, backlog practices |
About¶
This knowledge base is maintained by Victoria Crosby (@vcrosby22), a Technical Project Manager who uses Cursor daily and shares what works.
Licensed under CC BY 4.0 — share and adapt with attribution.