🧠 From Generic Assistant to MongoDB Expert

Out of the box, Cline is a great general-purpose coding assistant. Cline Skills are MongoDB-authored instruction sets that turn it into a domain expert—covering schema design, query optimization, search, and more. They’re already pre-installed in your workspace.

🔍 Step 1: Find Your Skills

The Skills panel lives behind the ⚖ scale icon in the Cline chat input area (not in Settings).

  1. Click the Cline icon in the VSCode sidebar to open the chat panel.
  2. In the chat input, click the ⚖ scale icon.
  3. You should see a list of MongoDB skills, including:
    • mongodb-schema-design
    • mongodb-natural-language-querying
    • mongodb-query-optimizer
    • mongodb-search-and-ai
    • mongodb-mcp-setup
    • mongodb-connection
    • atlas-stream-processing

⚠️ Note: If you don’t see any skills, reload the window: ☰ > View > Command Palette… > Developer: Reload Window.

🤖 Step 2: Try the Schema Design Skill

📋 Prompt: Start a new chat in Cline and paste this in:

Use the mongodb-schema-design skill. I’m migrating from a Postgres schema where bookings, listings, hosts, addresses, and reviews are five tables joined by foreign keys. I’m tempted to create five collections in MongoDB with ObjectId references that mimic the foreign keys. Why is that a mistake, and what should I actually do? Be specific about which tables to merge and which to keep separate, and why.

🎯 Step 3: Review the Response

The skill-powered answer should reference MongoDB-specific concepts by name. Look for:

  • Anti-patterns called out explicitly (e.g., excessive-lookups, unnecessary-collections)
  • Design patterns named (e.g., extended-reference, bucket, computed)
  • The embed-vs-reference framework applied per relationship (1:1, 1:few, 1:many, M:N)
  • A concrete recommendation, for example:
    • Embed address into listings (1:1, always accessed together)
    • Extended reference for host inside listings (cache name and picture_url, keep full host doc separate)
    • Reference for reviews (unbounded, can grow past the 16MB document limit)
    • Reference for bookings (independent lifecycle, queried separately)
  • The guiding principle quoted: “Data that is accessed together should be stored together.”

💡 The Takeaway

Each major section of this Arena has a matching skill:

  • 📊 Aggregations & slow queries → mongodb-query-optimizer
  • 🔎 Atlas Search & 🧠 Vector Search → mongodb-search-and-ai
  • ✏️ Writing queries from scratch → mongodb-natural-language-querying

Reach for the ⚖ scale icon whenever you start a new section. You’ll get answers tuned for MongoDB, not generic SQL-flavored guesses.

📚 Learn More

Ready for the next challenge?

Continue to: Explore Collection Structure