Portainer Container Compare Chart

Input schema for the renderer

Write JSON like this before rendering:

{
  "title": "Portainer comparison: graphite vs node-red",
  "subtitle": "12 live samples over ~55s via Portainer Docker stats",
  "items": [
    {
      "endpoint": "diskstation",
      "container": "graphite",
      "image": "graphiteapp/graphite-statsd:latest",
      "samples": [
        {
          "timestamp": "2026-04-05T15:00:00.000Z",
          "cpu_pct": 16.4,
          "mem_pct": 5.6,
          "mem_usage_bytes": 1234,
          "mem_limit_bytes": 5678,
          "rx_bytes": 100,
          "tx_bytes": 200,
          "pids": 42
        }
      ]
    }
  ]
}

Render command

Run the renderer script adjacent to this SKILL file:

bun ./render-portainer-container-compare.ts \
  --in /workspace/tmp/portainer-compare-input.json \
  --out-prefix /workspace/exports/portainer-compare

This writes:

• /workspace/exports/portainer-compare.svg
• /workspace/exports/portainer-compare.csv
• /workspace/exports/portainer-compare.json

Notes

• Keep data collection in the native portainer tool, not direct curl scripts.
• Use raw request for stats snapshots and unmodeled Docker-proxy paths.
• The renderer is local-only and intentionally dumb: it consumes normalized JSON and draws the chart.
• Use a bounded sampling window; this is for comparison, not long-lived streaming.
• Use the visual-design defaults: clean, minimal, visible axes, subtle grid, compact legend.