Level 300 credential route (draft): Open Certificate in Building with AI
Draft route for review. This page describes the Level 300 project and proposes using its marking sheet for the course credential. It does not award or promise a credential. The credential specification records the proposed route.
Anything marked Proposal is not decided yet. Bob Rapp and Ken Johnston, the review board co-chairs, decide it unless the page names Bob alone.
What you must do
Read the lessons and complete each module's quiz and workbook. For objectives 300.P-a and 300.P-b, enforce a hold for one system in the course sandbox, using the sample policy or your share-safe 200.P policy mapped to synthetic actions. Dry-run it on the synthetic log without changing any real system. Keep paper results separate from sandbox traces. Pass the readiness check, get a second person's review, and bring the share-safe evidence to a host.
Proposal, needs the review board co-chairs’ (Bob and Ken) decision: make the quizzes and workbooks practice only, not a credential requirement. The current credential specification calls for completing them.
The rubric: how your project is marked
| Check | Meets when |
|---|---|
| Holder and expiry | Every action has a named holder, approval ID and expiry, with an alert and escalation route |
| Authority table | Places every action under alone, approver or never, denies unlisted actions, and writes in the judgments |
| Out-of-band confirmation | Every irreversible action needs confirmation through a channel the requester does not control |
| Budget | Counts per reviewer and system-wide, by hour, and says what happens when the budget is exceeded |
| Stop | Fails closed if the holder, approval, record or monitor is missing and says when each condition becomes knowable |
| Receipts | Gives what happened, holder, approval, result and undo through references only |
| Coverage | States the unit, scope and window; independently counts executions with their own allow receipt and lists runs after refusal separately, or says it cannot count yet |
| Dry run | Completes all four runs on paper and in the sandbox; matches receipts to synthetic logs and traces, names gaps and untested conditions, and keeps paper and sandbox evidence apart |
A worked example
In Big Widget's made-up warranty-claims case, the dry run tests both the per-reviewer and system-wide budgets. A missing approval record is a separate run. Calling a stop condition tested requires an actual trace for it; writing down what ought to happen is paper evidence.
How a host marks it
The readiness questions are auto-graded before human marking; they are not the project mark. A second person checks the hold and receipts and signs and dates their findings. A host at the Foundation's Thursday call checks the hold, readiness result, synthetic logs and sandbox traces. Paper evidence is marked as paper evidence; the enforced part of the objective is marked from sandbox traces. You do not have to host a call.
If it is "not yet": revise and resubmit
Correct the hold or evidence, repeat the affected runs, and keep written-but-untested conditions separate from observed results. No separate credential resubmission or appeals process is established here.
What the certificate would say, and what it would not
The proposed claim is limited to a host marking this hold and dry run against the rubric. It would not certify a real system as safe or lawful or confer a degree, vendor credential, or professional certification.
The certificate would carry the lines "Free and open. Every source cited. Human-reviewed." and "Earned, never bought."