Skip to content

How delivery would work

Check that the plan preserves what matters

The cancellation plan adds validation and persistence. It gives no task for preserving the shipped-order guard. The reviewer sees which criterion has no task and returns the plan for revision.

Requirements traceability from tasks to tests

Where verification sits

See how criteria constrain a coding plan and how the team distinguishes proposed checks from results tied to a source version.

Criteria are written in Define, covered by tasks in Plan and checked in Verify. Each link is a record a reviewer can open.

  1. Understand

    Read the twin around the area you want to change: paths, rules, data and gaps.

    AI drafts · you read

  2. Define

    Write the use case, stories and acceptance criteria against the current model.

    AI drafts · lead edits

    Approve requirement
  3. Plan

    Propose files, symbols, tasks and the impact map. Every criterion needs a task.

    AI proposes · dev reviews

    Approve plan
  4. Implement

    The coding agent gets an approved, version-pinned brief and stops if inputs change.

    Agent writes · dev guides

    Review and merge
  5. Verify

    Test cases trace back to criteria. Results name the build they ran against.

    QA reviews · tests run

    Accept results
  6. Refresh

    The merge becomes a new commit. The twin rebuilds and old approvals are rechecked.

    Automatic

Every push starts the loop again from a fresh twin. Approvals built on the old code are flagged, not silently kept.

Walkthrough

Step by step.

  1. From criterion to task

    Map each requirement to the work that implements or preserves it. A criterion with no covering task is a planning gap. A task with weak source support also needs a review decision before it can count as a complete plan.

  2. From proposed check to observed result

    A configured test case describes the expected outcome. A result shows what ran, under which inputs, and against which version. Show incomplete runs, skipped cases and stale results by name.

  3. Read approvals and delivery evidence independently

    Review of a plan, QA review of cases, execution results, code review and merge answer different questions. Distinct records preserve who decided what and help a later reader see the behaviour a release actually exercised.

What a reviewer can actually conclude

ArtifactWhat it supportsStill needed
Criterion mapped to a taskThe plan accounts for the shipped-order ruleThe task and check still need execution
Passed fixture assertionThe guard resisted that input in the local fixtureProduction compatibility and database effects
Skipped client checkThe required client version was unavailableA supported older client or explicit rollout decision