Dashboard

Data Storage

~/dashboard/
├── registry.json           # Dashboard index
├── {name}/
│   ├── config.json         # Layout, widgets
│   ├── data.json           # Current data
│   └── index.html          # Dashboard page

Create on first use: mkdir -p ~/dashboard

Scope

This skill:

• ✅ Generates static HTML dashboards
• ✅ Creates fetch scripts user can run
• ✅ Stores dashboards in ~/dashboard/

User-driven model:

• User specifies data sources
• User provides API credentials via environment
• User runs fetch scripts (cron or manual)
• Skill generates HTML and fetch scripts

This skill does NOT:

• ❌ Access credentials without user providing them
• ❌ Run automated fetches (user's cron runs scripts)
• ❌ Scrape services without user consent

Quick Reference

Core Rules

1. User Provides Data

When creating a dashboard:

User: "Dashboard for my Stripe revenue"
Agent: "I'll create a fetch script. Set STRIPE_API_KEY 
        in your environment, then run the script."
→ Generates: ~/dashboard/stripe/fetch.sh
→ User adds to cron: */15 * * * * ~/dashboard/stripe/fetch.sh

2. Architecture

[User's Cron] → [fetch.sh] → [data.json] → [index.html]
                    ↓
              Uses $API_KEY from env

Agent generates scripts. User runs them.

3. Fetch Script Template

#!/bin/bash
# Requires: STRIPE_API_KEY in environment
curl -s -u "$STRIPE_API_KEY:" \
  https://api.stripe.com/v1/balance \
  | jq '.' > ~/dashboard/stripe/data.json

4. Visual QA (Before Delivery)

• Open in browser, take screenshot
• Check: no overlap, readable fonts (≥14px), good contrast
• If issues → fix → repeat
• Only deliver after visual validation

5. Design Defaults

6. Security

• Credentials via env vars, never in files
• Dashboards on 127.0.0.1 by default
• No PII in displayed data
• User adds auth if exposing to network