Skip to content
Get 12 months free access for taking part in the research as an early adopter.

Verification

From activity to evidence.

Activity is what a tool recorded. Evidence is activity that a second, independent source confirms. Only evidence reaches a profile, as counts and levels.

Checking a record

Sample
  • QR code

    Scan it, see the live record

  • Signed evidence

    Ed25519, public keys published

  • Timestamps

    Every item dated at its source

  • API

    One call from your own systems

How a claim is built

Intent → Behaviour → Evidence

Three stages, each reading only the one before it. That layering is the audit trail.

  • Intent

    What was set out to be done: a piece of work, logged as it finished, as one neutral line and a duration.

  • Behaviour

    What the record shows happened: sessions, practices, and how the work divided across domains.

  • Evidence

    What an independent source confirms: a repository, a CI run, a deploy, things a person cannot write about themselves.

Three strengths of evidence

Not every observation is a claim.

Every figure on a profile carries how it was confirmed, and the three strengths are kept apart.

  • Observed locally

    Logged by the person's own tools. Counts toward their private mirror; never a level on its own.

  • Corroborated across sources

    Two independent sources agree on the same stretch of work. This is where hours become verified hours.

  • Independently supported

    Confirmed by a source the person does not control (a CI run, a production deploy) and validated as a measure before it may be published.

What can become a claim

  • Hours that two independent sources agree on.
  • A practice observed in the record, within a stated window.
  • A level whose every gate is met, and named as insufficient when one is not.

What cannot

  • Anything a person can only say about themselves.
  • What someone is: no traits, no types, no work style on an employer surface.
  • Content of any kind, prompts, code, documents, client names.

Verify AI skills before you hire.

See the work behind a claim as hours, observed skills and evidence strength. Never a candidate's prompts, code or clients.