Full Feature

Full Feature Recipe

Step-by-step process for building features that span all layers of the Composable Stack: Data (PostgreSQL/NocoDB), Logic (n8n/Trigger.dev), and Interface (NocoBase/NocoDB).

Feature Development Process

Follow these steps in order for every new feature.

Step 1: Design Data Model (Data Layer)

1. Define tables in PostgreSQL using snake_case naming
2. Include standard fields: id, created_at, updated_at
3. Define relations (foreign keys, junction tables)
4. Create the tables using the postgresql-external-dev plugin skills
5. Alternatively, create tables directly using PostgreSQL MCP:

Tool: mcp__postgresql-mcp__execute_sql
Input: { "sql": "CREATE TABLE orders (id SERIAL PRIMARY KEY, title TEXT NOT NULL, status TEXT DEFAULT 'pending', created_at TIMESTAMPTZ DEFAULT now(), updated_at TIMESTAMPTZ DEFAULT now())" }

6. Verify schema with PostgreSQL MCP:

Tool: mcp__postgresql-mcp__list_tables
Input: { "table_names": "orders" }

Verify the tables also appear in NocoDB:

Tool: mcp__nocodb__getTablesList

Step 2: Set Up NocoDB Views (Data Layer)

1. Create appropriate NocoDB views (Grid, Form, Gallery, Kanban)
2. Configure field visibility and ordering
3. Set up filters for common queries
4. Create shared views if external access is needed

Step 3: Build NocoBase Interface (Interface Layer)

Always build on the NocoBase dev instance first (${NOCOBASE_DEV_URL} / ${NOCOBASE_DEV_API_KEY}, also exposed through the nocobase-dev MCP server). The dev instance is a sandbox for new tables, fields, menus, pages, blocks, and UX — safe to break, fast to iterate.

1. Create or sync NocoBase collections from the PostgreSQL tables on the dev instance
2. Design form blocks for data entry
3. Design table blocks for data display
4. Add action buttons for triggering workflows
5. Configure field permissions and validation rules
6. Once validated, promote the schema and UI to the prod instance (${NOCOBASE_URL}) via export/import or migration

Step 4: Create Automation Logic (Logic Layer)

Choose the appropriate service:

For n8n workflows (visual, short-running):

1. Create webhook trigger or schedule trigger
2. Build data transformation nodes
3. Add NocoDB read/write operations
4. Configure error handling
5. Name: [Domain] - [Action] - [Trigger]

For Trigger.dev tasks (durable, long-running):

1. Create task file in src/trigger/
2. Define payload type and retry config
3. Implement business logic
4. Add NocoDB status updates
5. Deploy with npx trigger.dev deploy

Step 5: Connect the Layers

Wire up the integration points:

1. Data → Logic: NocoDB webhook → n8n/Trigger.dev
2. Interface → Logic: NocoBase button/workflow → n8n webhook
3. Logic → Data: n8n/Trigger.dev → NocoDB MCP (read/write records)
4. Logic → Data (direct): n8n HTTP Request → PostgREST API (REST CRUD)
5. Logic → Data (SQL): Trigger.dev/n8n → PostgreSQL MCP execute_sql (complex queries)
6. Logic → Interface: n8n → NocoBase API (update UI state)

Step 6: Verify End-to-End

1. Create a test record through NocoBase UI or NocoDB
2. Verify the automation triggers
3. Check the result in NocoDB
4. Verify status updates in the interface

Feature Template

Use full-feature/template.md as a structured checklist for each new feature. Copy and fill in the placeholders.

Example: Order Processing Feature

Best Practices

• Start with the data model — it's the foundation for everything
• Use NocoDB views for operational data access, NocoBase for admin UI
• Keep n8n workflows focused — one workflow per business process
• Use Trigger.dev for anything that needs retries or takes >30 seconds
• Document webhook URLs and task IDs in the project config
• Test each layer independently before wiring them together
• Include a status field on records that participate in workflows