Skip to content

Tutorial: Create a Personalized App from the Terminal

Build a fully structured app using natural conversation with an AI coding assistant. No dashboard clicks, no REST calls, no schema design.

Prerequisites

  • A Nexo account
  • A developer key (Dashboard → Profile → Developer Access)
  • Claude Code installed (or any MCP client)

Step 1: Connect

export NEXO_DEVELOPER_KEY=nexo_uak_...
export NEXO_BASE_URL=http://localhost:8000
claude mcp add --scope project --transport http nexo-mcp \
  "${NEXO_BASE_URL}/mcp" \
  -H "X-Api-Key: ${NEXO_DEVELOPER_KEY}"

Or run bash scripts/connect-mcp.sh from the repo root.

Verify the connection inside Claude Code with /mcp — you should see nexo-mcp listed with 11+ tools.

Step 2: Create an app

Tell the agent what you want:

"Create an expense tracker for a shared apartment. We split rent, utilities, groceries, and eating out."

What happens behind the scenes:

  1. The agent calls micro_apps__plan_app with your prompt. Nexo returns a structured template:

    {
  "status": "ok",
  "source": "template_engine",
  "template": {
    "name": "Shared Apartment Expenses",
    "archetype": "shared_expenses",
    "tables": [
      {
        "key": "expenses",
        "label": "Expenses",
        "fields": [
          { "key": "description", "label": "Description", "type": "text", "required": true },
          { "key": "amount", "label": "Amount", "type": "number", "required": true },
          { "key": "category", "label": "Category", "type": "select", "options": ["rent", "utilities", "groceries", "eating_out", "other"] },
          { "key": "paid_by", "label": "Paid by", "type": "text", "required": true },
          { "key": "date", "label": "Date", "type": "date", "required": true },
          { "key": "split", "label": "Split", "type": "number" }
        ]
      },
      {
        "key": "people",
        "label": "Roommates",
        "fields": [
          { "key": "name", "label": "Name", "type": "text", "required": true }
        ]
      }
    ],
    "views": [
      { "key": "all_expenses", "type": "list", "table_key": "expenses" },
      { "key": "by_category", "type": "cards", "table_key": "expenses" },
      { "key": "summary", "type": "summary", "table_key": "expenses" }
    ],
    "seed_records": [
      { "table_key": "expenses", "values": { "description": "April rent", "amount": 1200, "category": "rent", "paid_by": "Alice", "date": "2026-04-01", "split": 3 } }
    ]
  }
}

  2. The agent calls micro_apps__provision_app with the template. Nexo creates the app:

    {
  "status": "ok",
  "app": { "id": "a1b2c3d4-...", "name": "Shared Apartment Expenses" },
  "tables_created": 2,
  "fields_created": 7,
  "views_created": 3,
  "records_created": 1
}

Your app now exists with 2 tables, 7 fields, 3 views, and a seed record.

Step 3: Add data

"Add: $120 groceries, paid by Alice, split 3 ways"

The agent calls micro_apps__add_record:

{
  "app_id": "a1b2c3d4-...",
  "table_key": "expenses",
  "values": {
    "description": "Groceries",
    "amount": 120,
    "category": "groceries",
    "paid_by": "Alice",
    "date": "2026-04-12",
    "split": 3
  }
}

Response:

{
  "status": "ok",
  "record": {
    "id": "rec-uuid-...",
    "values": { "description": "Groceries", "amount": 120, "category": "groceries", "paid_by": "Alice", "date": "2026-04-12", "split": 3 }
  }
}

Step 4: Evolve the schema

"Add a 'settled' checkbox field to track which expenses have been paid back"

The agent calls micro_apps__plan_operation:

{
  "prompt": "Add a settled checkbox field to track which expenses have been paid back",
  "app_id": "a1b2c3d4-..."
}

Response:

{
  "status": "ok",
  "source": "template_engine",
  "operation": {
    "kind": "add_field",
    "app_id": "a1b2c3d4-...",
    "table_key": "expenses",
    "field": {
      "key": "settled",
      "label": "Settled",
      "type": "boolean",
      "required": false
    }
  }
}

Then the agent calls micro_apps__apply_operation with the operation object:

{
  "status": "ok",
  "kind": "add_field",
  "app_id": "a1b2c3d4-...",
  "table_id": "tbl-uuid-...",
  "resource_id": "fld-uuid-...",
  "resource_type": "field"
}

The new field is live. Existing records get settled: null by default.

Step 5: Query your data

"Show me all unsettled expenses over $50"

The agent calls micro_apps__query_app:

{
  "app_id": "a1b2c3d4-...",
  "table_key": "expenses",
  "filters": [
    { "field": "amount", "op": "gte", "value": 50 },
    { "field": "settled", "op": "neq", "value": true }
  ],
  "sort": [{ "field": "amount", "direction": "desc" }]
}

Response:

{
  "status": "ok",
  "records": [
    { "id": "rec-1", "values": { "description": "April rent", "amount": 1200, "category": "rent", "paid_by": "Alice", "settled": null } },
    { "id": "rec-2", "values": { "description": "Groceries", "amount": 120, "category": "groceries", "paid_by": "Alice", "settled": null } }
  ],
  "total": 2
}

Step 6: Get the full picture

"Show me a summary of everything"

The agent calls micro_apps__get_context:

{
  "status": "ok",
  "markdown": "## Shared Apartment Expenses\nExpense tracker | 2 records | Last updated: Apr 12\n### expenses (7 fields, 2 records)\nFields: description (text, required), amount (number, required), category (select), paid_by (text, required), date (date, required), split (number), settled (boolean)\nRecent:\n- Apr 12: Groceries, $120.00, groceries (Alice)\n- Apr 1: April rent, $1200.00, rent (Alice)\n### people (1 field, 0 records)\nFields: name (text, required)",
  "app_count": 1,
  "total_records": 2
}

This compact markdown is designed for LLM context injection — the agent can use it to understand your data before answering follow-up questions.

What you built

In 6 conversational turns:

  • A 2-table app (expenses + roommates) with typed fields
  • 3 views (list, cards, summary)
  • Seed data from the template
  • A custom boolean field added after creation
  • Filtered and sorted queries
  • A full-context markdown summary

The same app is visible in the Nexo dashboard at https://nexo.luzia.com/dashboard and can be accessed via the REST API.

Using other MCP clients

The same workflow works in any MCP client. The key steps are always: 1. Connect to ${NEXO_BASE_URL}/mcp with your developer key 2. Ask the agent to plan an app 3. Review and provision 4. Evolve with operations

Claude Desktop: Add the MCP server in ~/Library/Application Support/Claude/claude_desktop_config.json (see connection guide). Then say "Create a meal planner" in a new conversation.

Cursor: Add the MCP server in Settings -> MCP (see connection guide). Use the chat panel to describe apps.

Codex / Windsurf: Connect using the same HTTP transport URL and header. The tool sequence (plan_app -> provision_app -> show_app -> plan_operation -> apply_operation) is identical across all clients.

Tip: For richer apps, reference the available primitives in your prompt: "Create a training plan with a HeroCard daily view, milestones, and a progress_ratio metric."

Next steps