Good Old Search

Build precise search queries with exact phrases, site filters, exclusions, date ranges, filetypes, and one-click links across search engines and vertical search targets.

This is not a search engine, metasearch aggregator, scraper, or API proxy. It is a local static query builder that opens the search engine you choose.

The app has five surfaces:

• Simple: the everyday exact-search builder at index.html.
• Advanced Search Lab: a pro operator builder at advanced.html for Google dorks, filetype searches, site/path filters, title/URL/body operators, date ranges, presets, and raw operator lines.
• Shodan: a Shodan query builder and search helper at shodan.html for banner terms, org/net scopes, ports, TLS certificate filters, HTTP metadata, screenshots, and defensive exposure review.
• crt.sh: a crt.sh query builder and search helper at crtsh.html for certificate transparency searches, wildcard domains, expired certificate cleanup, deduped results, JSON exports, and certificate ID pivots.
• Wayback: a Wayback query builder and search helper at wayback.html for CDX API URLs, match types, date ranges, status codes, MIME filters, collapsed captures, and archive research.

Supported Targets

Web:

• Google (with udm=14 to skip the AI Overview)
• Google Verbatim
• Brave
• Kagi
• DuckDuckGo
• Bing
• Marginalia (small-web, non-commercial index)
• Mojeek (independent crawler)

Technical and forums:

• Hacker News
• GitHub Issues
• GitHub Code
• Stack Overflow
• Reddit via Google

Research and archives:

• arXiv
• Google Scholar
• Wayback Machine (Internet Archive web captures)

Recipes

• Exact: quote the whole input.
• Old Google: quote the input and prioritize stricter web searches.
• Technical Error: quote the full error and recommend GitHub Issues, Stack Overflow, HN, and code search.
• Forums: prioritize human discussion surfaces.
• Old Web: add a before: date, with quick presets for 2022, 2020, 2016, and 2010.
• PDFs / Docs: add filetype:pdf.
• Exclude Sludge: apply an editable local blocklist for -site: filters and recurring junk terms.

Local Development

npm install
npm run dev

The dev command builds the static app once and serves dist at http://127.0.0.1:4173. There is no HMR. Re-run npm run dev (or npm run build && npm run preview) after changes. The build compiles the TypeScript app, copies static assets, and emits sitemap/robots files.

Build

npm run build

Output goes to dist/. The app can also be opened directly from dist/index.html.

Set PUBLIC_SITE_URL at build time for correct canonical URLs and sitemap entries. The default is https://goodoldsearch.com.

PUBLIC_SITE_URL=https://goodoldsearch.com npm run build

Docker / Coolify

The included Dockerfile builds the app, then serves dist with the existing Node static server.

Coolify settings:

• Build pack: Dockerfile.
• Exposed port: 3000.
• Build arg or environment during build: PUBLIC_SITE_URL=https://goodoldsearch.com.
• Runtime env: Coolify can provide PORT; otherwise the image defaults to 3000.

Tests And Smoke

npm test
npm run smoke
npm run smoke:live
npm run smoke:live:strict

• npm test builds the app, runs the single-process assertion suite, and runs query-type smoke checks.
• npm run smoke verifies every supported engine URL builder, host, engine-specific query syntax, key recipe behavior, and share URL round trip.
• npm run smoke:live also tries live reachability for every target and reports external warnings without failing the structural smoke.
• npm run smoke:live:strict fails on external reachability warnings too. Some engines may rate-limit, redirect, or bot-block automated requests even when browser links work.

File Structure

src/
  main.ts
  styles.css
  lib/
    queryModel.ts
    queryBuilder.ts
    recipes.ts
    shareUrl.ts
    blocklist.ts
    bookmarklet.ts
    engines/
      index.ts
      google.ts
      googleVerbatim.ts
      bing.ts
      brave.ts
      duckduckgo.ts
      kagi.ts
      hn.ts
      redditGoogle.ts
      stackoverflow.ts
      githubIssues.ts
      githubCode.ts
      arxiv.ts
      utils.ts
tests/
  query.test.mjs
scripts/
  build-seo-files.mjs
  clean-dist.mjs
  copy-static.mjs
  serve-dist.mjs
  smoke-engines.mjs

License

MIT. See LICENSE.

Caveats

• Google can match quoted terms in hidden page text, URLs, or stale indexed content.
• DuckDuckGo may show related results if exact matches are sparse.
• Bing may ignore punctuation or limit how many terms affect results.
• Kagi may require login.
• Vertical searches ignore web-only operators where their engine does not support them.