Public Document Finder — Domain & Company Files

Find publicly discoverable documents and data files for a company, domain, or keyword — PDFs, Office, CSV/JSON, archives, and more. No logins and no vendor API keys — discovery runs through HTTP automation only.

Use it for competitive research, vendor diligence, journalism, and authorized recon when you need a structured list of public file URLs (not private contents behind auth).

What you get

Each dataset row is one unique document URL with:

Field	Meaning
url	Direct or share URL
title	Best-effort title
fileType	Extension hint (pdf, docx, …)
hostClass	website, google_drive, amazon_s3, wayback, github, …
channel	Which discovery path found it
interestScore	0–100 heuristic OSINT triage score (cloud shares, budgets, decks, etc. rank higher; robots.txt-style noise is skipped)
snippet	Search snippet when available

How to run

1. Enter a domain (mozilla.org) and/or keyword (annual report).
2. Keep Maximum documents at 5 for a cheap first run.
3. Click Start. Export JSON/CSV or call the API.

Google / Bing filetype search

Builds grouped filetype: OR queries across documents, sheets, decks, archives, and data/config files, then collects matching public URLs from search HTML.

Google Drive & Google Docs (public)

Searches for indexed public Drive/Docs share links tied to your brand or domain. Only links search engines already expose; nothing behind a Google login.

Amazon S3 / open cloud objects

Looks for publicly indexed object URLs and share pages (S3-style hosts and similar) associated with your query. Useful when marketing sites or blogs link to files on cloud storage.

Wayback Machine / Internet Archive

Queries Wayback CDX for historical captures whose URLs end in OSINT-relevant extensions (PDF, Office, CSV, ZIP, JSON, …) on the target domain tree (matchType=domain).

GitHub public assets

Finds public GitHub-hosted files related to the brand via search pivots (no GitHub API token).

Website sitemaps

Reads robots.txt sitemap pointers and walks sitemap indexes for document / data / archive file URLs published by the site itself.

Well-known / agent-discovery files

Probes /llms.txt, /llms-full.txt, /AGENTS.md, /.well-known/security.txt, and related public discovery files. Rows get an interestScore plus optional discoveryFlags (e.g. possibleAgentInstruction) when the text looks like agent-facing policy or prompt-injection shaped language — useful for AI-security OSINT without logging in.

Subdomains

When Include subdomains is on (default), the Actor discovers public hostnames under your apex domain via certificate transparency and free host-search pages (no API keys), then extends site: search queries, sitemaps, and shallow crawls to those hosts (capped by Maximum subdomains). Common Crawl and Wayback already query the whole domain tree via wildcard / matchType=domain.

Sources

Additional channels used when enabled:

• Common Crawl public index (URL extension filter across the default OSINT file-type set)
• Shallow same-site crawl of docs/research/download-style paths
• Dropbox / OneDrive / SharePoint / Box public links (search-indexed only)
• Azure Blob / GCS style hosts when they appear in search
• SlideShare / SpeakerDeck when they match the query

Turn channels on/off with the Discovery channels input: search, commonCrawl, wayback, sitemap, siteCrawl, cloudShares, wellKnown.

Input tips

• Prefer a hostname (who.int) over a long marketing URL.
• Add an Extra keyword to narrow (strategy, transparency, pricing).
• Increase Maximum documents after a successful 5-row smoke.
• Enable Apify Proxy if search HTML is empty from datacenter IPs.

Output & pricing

Results land in the default dataset. Pricing is pay per result on the Store Pricing tab (Free vs paid plans differ). The Actor stops early if your run charge limit is reached.

Limitations

• Only public / indexed material — not authenticated portals, paywalls, or private Drive folders.
• Search and archive sources can rate-limit or time out; try proxy or fewer channels.
• Titles/snippets are best-effort; always open the URL to verify.
• Unofficial tool — not affiliated with Google, Amazon, Microsoft, Internet Archive, or any site you query.

Related

Pair with a PDF text extractor Actor if you already have URLs and need full text. This Actor finds public document URLs; it does not OCR or parse file bodies.