Run, Grade, Repair, and Record

What you will complete

A passing run log or an honest blocked result after focused repair cycles.

Make the operating decision

Run the same case and grade every criterion pass or fail. Diagnose whether the problem comes from instructions, deterministic code, source data, permissions, or an unresolved product choice. Preserve the failed output and version before editing the artifact.

Make one focused correction, rerun affected cases, and append the new result. Do not weaken the expected result to manufacture a pass. After three unsuccessful repair cycles, stop and request the missing decision or reduce scope. A blocked build with clear evidence is more reliable than a green label built on untested assumptions.

Inspect the artifact itself. Confirm source row IDs, missing values, output location, and lack of external effects. If an independent fresh-context test is permitted, give it the normal inputs rather than the desired answer or suspected bug.

Worked fictional example: Harbourlight Home Services

Harbourlight's first run omits unknown-owner rows because the prompt treats blank as no exception. The team records the failure, corrects the explicit unknown rule, and reruns the same case plus the held-back case. The run log retains both versions and verdicts.

Harbourlight Home Services and every estimate record in this course are fictional. They demonstrate the method and do not represent a Flowgrammer client, a deployed system, or measured savings.

Complete workbook section 7

Use the evidence available for your own bounded job. Write unknown when the evidence is missing, and record the person or action that can resolve it.

1. Run the representative case and grade every outcome.
2. Classify each failure by layer.
3. Make one narrow repair and rerun affected cases.
4. Stop after three failed cycles and document the unresolved decision.

Open the course workbook PDF or read the workbook details.

Critical gate before continuing

• Failed evidence remains available.
• Expected results do not move to match output.
• The held-back case passes independently.
• No untested criterion is marked pass.

If a gate fails, repair the current section, narrow the scope, leave the route manual, or record a blocked or stop decision. Continuing is not the only successful learner action.

Common failure modes

• Expanding beyond a passing run log or an honest blocked result after focused repair cycles. before the current artifact can be graded.
• Turning a missing value, unavailable source, or blocked integration into a confident conclusion.
• Treating a prompt instruction as proof that the effective tool or permission boundary works.
• Marking a manual, simulated, or untested route as live.

Check your application