sanity-log-analyzer

Analyze Sanity request logs and generate an offline report to help triage bandwidth usage, expensive queries, and oversized images. No server required.

Why this exists

Sanity's request logs contain a huge amount of operational data, but they're difficult to inspect manually. This package turns raw NDJSON logs into visual reports that surface common fixes for out-of-control bandwidth.

Features

• Interactive self-contained HTML report
• Structured Markdown export for analysis with LLMs
• Detects common bandwidth hogs
• Highlights oversized image requests
• Identifies expensive GROQ queries
• GROQ query viewer with syntax highlighting
• Streams multi-GB NDJSON logs without loading them into memory
• Analysis happens entirely on your machine. Request logs are never uploaded or transmitted

Requirements

• Node.js >= 22

Install

npm install sanity-log-analyzer
bun install sanity-log-analyzer
yarn add sanity-log-analyzer
pnpm add sanity-log-analyzer

CLI

npx sanity-log-analyzer input.ndjson output.html --open

⠴ Processed 1.2 GB / 1.2 GB (98.9%) — 1,264,879 entries
Wrote output.html in 6.6s (1,284,676 requests).

HTML report

sanity-log-analyzer <input.ndjson|.ndjson.gz> <output.html> [--config config.json] [--open]

• Streams large NDJSON files without loading everything into memory
• Writes a self-contained HTML file (inline CSS, no server required)
• --config — optional JSON file to customize title, sections, and top-N
• --open — open the report in your default browser after writing

Config

// Minimal config:

{
  "title": "Production Report",
  "topN": 50,
  "sections": {
    "images": true,
    "referrers": false
  }
}

Optional JSON file passed via --config. Merges with defaults.

• title — report heading
• topN — max rows per breakdown table (default: 50)
• sections — toggle individual report sections on/off (images, files, queries, responseStatuses, hourlyBandwidth, dailyBandwidth, referrers, userAgents, ips)

Programmatic API

import { analyzeLog, writeHtmlReport } from "sanity-log-analyzer";

const report = await analyzeLog("logs.ndjson", {
  config: { title: "My Report" },
  onProgress: (p) => console.log(p.percent),
});

await writeHtmlReport(report, "report.html");

• analyzeLog(inputPath, options?) — returns ReportData
• writeHtmlReport(report, outputPath) — writes self-contained HTML
• generateMarkdown(report, options?) — returns the report markdown
• writeMarkdownReport(report, outputPath, options?) — writes markdown to disk
• markdownReportFilename(report) — derives the markdown filename from the report title
• resolveReportConfig(partial) / loadReportConfig(path) — config helpers

Markdown export

import { analyzeLog, generateMarkdown, writeMarkdownReport } from "sanity-log-analyzer";

const report = await analyzeLog("logs.ndjson");

const markdown = generateMarkdown(report);

await writeMarkdownReport(report, "report.md");

The HTML report includes a Download markdownM button that downloads a .md file.

Input format

• Newline-delimited JSON (.ndjson) or gzip-compressed (.ndjson.gz)
• One Sanity API request log entry per line

Roadmap

• [ ] Support Sanity request tagging via the ?tag=... param
• [ ] Rewrite parser in Rust