Lens

Instant dashboards from Claude Code. One prompt. Under a minute. Done.

Point Claude at your data -- PostgreSQL, CSV, Excel, Parquet -- and get a production-ready analytics dashboard. No frontend code. No setup. No cleanup needed.

pip install mindcase-lens
claude mcp add lens -- uvx mindcase-lens-mcp

Then tell Claude:

"Build me a dashboard from my database at postgresql://localhost:5432/mydb"

That's it. Claude introspects your schema, generates the config, and serves a full dashboard with KPIs, charts, tables, filters, and theming.

See It in Action

You: "I have a folder of CSV files in ./data/ — build me a dashboard"

Claude: Introspecting ./data/...
        Found 3 tables: sales (10K rows), customers (500 rows), products (200 rows)
        Generating dashboard config...
        Dashboard running at http://localhost:8080

Under a minute. Every time.

What You Get

• KPIs with trend indicators and % change
• Charts -- bar, line, area, pie, donut, horizontal bar (with pagination for dense data)
• Data tables -- sortable, paginated, conditional formatting, CSV export
• Filters -- dropdowns (single/multi with search), date range, text, number range, toggle
• Theming -- light/dark mode, color themes, fonts, corner radius. All live-preview.
• Sidebar navigation -- auto-generated, with logo support

Any Data Source

# One file
connection: "sales.csv"

# A whole folder — each file becomes a table
connection: "./data/"

# Multiple sources together
connection:
  - "sales.csv"
  - "customers.xlsx"
  - "postgresql://localhost:5432/mydb"

Why Lens?

Lens exists because every other dashboarding tool requires you to either build a frontend, learn a UI, or write code. With Lens, you describe what you want to Claude, and it's done.

For Power Users

Everything Claude generates is a YAML config file. You can edit it directly:

app:
  title: "Sales Dashboard"
  database:
    connection: "postgresql://localhost:5432/mydb"
  pages:
    - id: sales
      name: Sales
      icon: bar-chart
      filters:
        - id: region
          type: dropdown
          label: Region
          query: "SELECT DISTINCT region FROM sales"
          all: true
      rows:
        - height: small
          items:
            - type: kpi
              title: Revenue
              query: "SELECT SUM(amount) as current_value FROM sales WHERE (:region = 'ALL' OR region = :region)"
              prefix: "$"
              compact: true
        - height: medium
          items:
            - type: chart
              chart_type: bar
              title: Revenue by Month
              query: "SELECT month, SUM(amount) as revenue FROM sales GROUP BY month"
              x: month
              y: revenue
              y_format: currency

Full YAML reference: llms-full.txt

CLI

lens serve dashboard.yaml              # start dashboard
lens serve dashboard.yaml --port 3000  # custom port
lens validate dashboard.yaml           # validate config

Python API

from lens import Lens

app = Lens("dashboard.yaml")
app.serve(port=8080)

Privacy & Security

• All database connections are read-only by default
• SQL queries are parameterized -- no injection risk
• Raw SQL never reaches the browser
• No telemetry, no external calls, no data leaves your machine

Also Available

• Lens MCP Server -- the Claude integration (pip install mindcase-lens-mcp)

License

MIT -- Built by Mindcase