| name | executing-plans |
| description | Use when executing an approved APIVR implementation plan sequentially in the current session without subagents, while preserving TDD, review, ledger, and release gates. |
| activation | Activate when the description trigger applies to the current task. |
| required_inputs | Task request, relevant repository context, constraints, and authority dependencies. |
| required_outputs | Skill-specific artifact, verification evidence, canonical verdict, and next action. |
| authority_dependencies | 00_start_here/SOURCE_OF_TRUTH.md; 10_governance/APIVR_EXECUTION_LIFECYCLE.md; 10_governance/source_of_truth/Elite_Build_Goals_v3.md. |
| evidence_requirements | Executed checks or an honest Unknown, Not Run, or Blocked state for every material claim. |
Executing Plans
Execute an approved plan one verifiable slice at a time. A plan authorizes only
the stated change, not improvisation around a failing command or vague scope.
Before the First Task
git status --short --branch
git log -1 --oneline
Confirm the branch, worktree, plan revision, target behavior, and rollback
point. Read all task dependencies before starting, then start only the first
independently testable slice.
Process
- Copy the exact task objective and acceptance check into a task receipt.
- Write or identify the focused failing test. Run it and capture Red evidence.
- Make the smallest production change that can make that test pass.
- Re-run the focused test, then its direct neighbors. Refactor only while
they remain Green.
- Inspect the diff for unplanned files, generated artifacts, and weakened
tests. If scope drift appears, stop and return to the plan owner.
- Commit or checkpoint only a coherent task boundary. Do not batch unrelated
“while here” cleanup into a verified slice.
Per-Task Receipt
Task and plan section:
Behavior and preserved behavior:
Red command and observed failure:
Green command and result:
Files changed:
Diff / scope audit:
Rollback point:
Evidence state and next step:
Worked Example
node --test scripts/tests/v2-registry.test.mjs
node scripts/generate-v2-metadata.mjs
node --test scripts/tests/v2-registry.test.mjs
git diff --check
git diff -- scripts/generate-v2-metadata.mjs runtime_adapters/
If the generator changes an unrelated manifest, that is a scope failure, not a
reason to update the plan silently.
Stop Conditions
- A required test fails for a different reason: repair the test setup first.
- A task needs new architecture, authority, credential, or user-file change:
return to
writing-plans.
- A checkpoint reveals a security or data-integrity risk: escalate APIVR tier.
When every task receipt is complete, use verification-before-completion, then
finishing-a-development-branch for the branch decision.