Make your learning visible

A credential records what your saved lab work demonstrates—not how many pages you visited. AI help is welcome. You keep control of what you share.

See my credential progress

Verified Core Lab Practitioner

Show that you can build a useful AI workflow, test its release, diagnose and repair failures, and adapt it to a changed task.

  • Release: fresh-user checks, denied access, fallback, restore and retest.
  • Capstone: task and boundary checks, reviewed outputs, handoff and value comparison.
  • Repair: evidence-linked diagnoses and verified repairs for each incident case.
  • Transfer: reviewed first and changed-context tasks with a current handoff.

These are the saved tasks in Core. The later spaced transfer recheck supports retention; it does not delay issuance.

Explore Core

Verified Agent Lab Practitioner

Demonstrate the current native Agent labs: operations, least authority, tool reliability, memory freshness, MCP trust and incident recovery. Each requires a current authenticated passing evaluation.

The integrated Agent task accepts current Core task evidence or the included prerequisite challenge. Labs assigned only to another specialization are not credential requirements.

Explore AI Agent Path

From practice to a private credential

  1. Open your progress to see the exact tasks and your saved results.
  2. Finish or revise the tasks. There is no separate exam, submission quota or retake fee.
  3. Choose your display name or a pseudonym and issue your signed credential privately.
  4. Download the signed record. Optionally create an unlisted link or publish a verification page.

A service problem does not count as a failed attempt. Your work stays saved; support can investigate an unexpected result. Issuance does not change your refund rights.

For someone checking a credential

Use the holder’s verification link to inspect the signed issuer, assessed task and criterion versions, evidence hashes, permitted assistance, attribution and current validity. A picture or completion count is not the verification record.

Anyone with an unlisted link can view it or forward it. An unavailable or withdrawn public page is not evidence of fraud. Validity, publication and access are separate.

What this does—and does not—prove

The record covers bounded lab performance, including fictional scenarios. It does not prove deployment to the learner’s own cloud account, professional expertise or performance in every workplace. Written reflections are not independently human-graded.

Identity assurance is verified control of the account email plus account continuity, not government-ID or employment verification. This is not accreditation, regulatory approval, an employment guarantee or 1EdTech certification.

Other specializations

Software Engineering, Security & Red Teaming, Evaluation & Reliability, Product Engineering, Privacy Engineering, and Voice & Conversational AI each need their own outcome-specific assessment—not a copied Core badge.

Their intended evidence is respectively a tested software change; an authorized attack-and-repair record; a defensible evaluation and reliability decision; a working product experiment; verified data-lifecycle and access controls; and a tested real-time conversation with interruption, consent and recovery. These specialized credentials are not included in the current Core and Agent credential definitions.

See the specialization labs