# Run a 10-change harness pilot

Blank worksheet for One engineering change, through a harness.

Source lesson: https://torque-and-tokens.the-tech-storyteller.com/radar/one-change-through-a-harness/

## Work unit / change request

[Complete with your own change request.]

## Part and configuration

[Complete with your own change request.]

## Pinned baseline / source IDs

[Complete with your own change request.]

## Unresolved version conflicts

[Complete with your own change request.]

## Tools and permissions (read / edit-in-branch / human-only)

[Complete with your own change request.]

## Dependency graph and joins

[Complete with your own change request.]

## Checks and acceptance criteria

[Complete with your own change request.]

## Independent reviewer

[Complete with your own change request.]

## Approval owner / evidence / permitted next action

[Complete with your own change request.]

## Decision record and retained evidence

[Complete with your own change request.]

## Rollback path

[Complete with your own change request.]

## Harness version and evaluation set

[Complete with your own change request.]

## Exercise

1. Pick 10 representative, closed historical change requests of one type, e.g. a bracket or a sensor change.
2. Freeze the source set for each: revisions, CAD configuration, requirements, test data.
3. Write one harness map: work unit, tools with permissions (read / edit-in-branch / human-only), dependency graph, checks with pass criteria, gate owners.
4. Run the changes with the harness and compare them with how they actually went, using the same acceptance rubric. Where you can, have a reviewer judge blind.
5. Track engineer hands-on time, elapsed queue time, first-pass review acceptance, evidence completeness, rework, escaped errors, human interventions and cost.
6. Write down scope and failures next to the result, and never relabel a faster retrieval step as end-to-end speed-up.

Expected deliverable: A one-page harness map plus a pilot scorecard for 10 changes: baseline vs harness on the same rubric, with scope and failures stated.