Trust rules
Recruiters get burned by tools that overclaim. These are the rules SourcingOS will not break, in the product and in the code.
No fake candidates
SourcingOS never invents people, companies, jobs, links, or contact data. If a search finds nothing, you see an honest empty state with suggestions for the next search, not filler results.
Public evidence only
Every record traces back to a public URL or to data you imported yourself. Source provenance stays attached to each evidence artifact so you can always answer the question: where did this come from?
Signals are not verification
Clearance language in a public bio is an unverified clearance breadcrumb. Open-to-work phrasing is a public signal. A matching name across two profiles is a possible match. None of these are treated as verified. Human verification is required before any of them count.
No silent merges
Source profiles stay separate until a recruiter reviews the identity signals and confirms they belong to the same person. There is no auto-merge at any confidence level.
No auto-outreach
SourcingOS drafts outreach angles grounded in the evidence you selected. It never sends anything. You review, edit, and send from your own tools.
No restricted-platform scraping
SourcingOS does not scrape LinkedIn, Indeed, ClearanceJobs, or paid resume databases, and it does not automate anything behind a login wall. X-Ray search strings are provided for you to run in your own browser and review yourself.
Fail closed
Private routes and write APIs require authentication. If the auth backend is unavailable, they refuse the request instead of quietly letting it through.
See it in practice
The methodology page walks through how these rules shape the actual workflow, and the data sources page lists exactly which public sources power search.