Most of the compliance writing on this site ends in “it depends”. This one mostly does not, and the reason is worth understanding: federal contract notices are not merely public, they are published because a rule requires agencies to publish them.
Why this answer is cleaner than usual
Two instruments do most of the work. FAR Part 5 requires agencies to publicise contract actions so that businesses can find them, and 17 U.S.C. § 105 places U.S. Government works outside copyright.
Compare that with Nextdoor, where the same request can return a business phone number and a resident's post about a local incident. Here the purpose of publication is the purpose you are using it for.
FAR Part 5: publication is the point
The Federal Acquisition Regulation requires agencies to publicise contract actions. The reason is competition: a government that buys without telling anybody gets fewer and worse offers.
So a tool that helps a business find opportunities it is eligible for is not working against the grain of the publication — it is the use case the publication exists to serve. That is a genuinely unusual position for a scraper to be in.
17 U.S.C. § 105 and reuse
U.S. Government works are not subject to copyright. That removes the question that complicates redistribution everywhere else: you are not licensing somebody's text.
It does not remove every consideration — a government work can incorporate third-party material, and “not copyrighted” is not the same as “cleared for any use”. But the default is markedly more permissive than for the commercial sites in this catalogue.
What is left is conduct
Three commitments, all of them behaviour rather than permission:
- robots.txt is fetched on every run and enforced in the request layer, so a disallowed path is never requested — not filtered afterwards, not requested.
- 4 concurrent requests with a one-second delay. SAM.gov is public infrastructure paid for by taxpayers, and an absent rate limit is not an invitation.
- No credential, no login, no evasion. The public search endpoint, used the way the search page uses it.
The documents that are public-facing and not public
This is the one place where care is genuinely required. Some attachments are marked restricted, export-controlled, deleted or non-public. Those are not returned at all — not the file, and not the URL, because publishing a URL to a controlled document publishes the document.
The check is an allowlist over six separate flags: an attachment is returned only if it is affirmatively public. A blocklist fails open when a new flag appears, and export control is not a place to fail open.
Personal data in a public record
Notices carry contracting officers' names, emails and sometimes phone numbers. That is a public record of an official function, published by the agency as the point of contact — which is about as defensible as personal data in a scraped dataset gets.
It is still personal data. A contracting officer publishing an email so vendors can ask about a solicitation has not consented to a marketing list, and the distinction between collecting and using applies here as it does everywhere else on this site.
A practical checklist
- Use the contact fields for the solicitation they belong to. Not for general outreach.
- Do not attempt to reach excluded attachments. If the Actor withheld a URL, that is the correct outcome.
- Attribute and link when you republish. No copyright does not mean no provenance.
- Take advice on export control if your work touches defence technical data. That regime is not about copyright and is not settled by anything on this page.
