AI SKILL · WORKING METHODS

Two-judgment handoff Skill

Check delivery quality separately from downstream readiness.

Trial example · 2026-11-17 · Not validated in a business pilot

When to use it

After an upstream deliverable is available, when the next process needs to assess acceptance and readiness. It organizes checks; it does not replace acceptance, authorization or external commitments by responsible people.

Inputs

Deliverable and version; agreed acceptance criteria; downstream permissions, resources and dependencies; owners on both sides; exceptions and evidence. A statement that work is complete is not proof of a successful handoff.

Steps

  1. Check the deliverable against each criterion: pass, fail or unknown, with evidence.
  2. Separately check downstream conditions: available, missing or unknown.
  3. Route quality issues to delivery/acceptance owners and readiness gaps to the next process or condition providers.
  4. Record actions, owners and review dates; let an authorized person decide whether to start.
  5. Record actual downstream adoption later: permission to start is not evidence of use.

Output format

Deliverable version:
Judgment A | Quality, evidence and acceptance owner:
Judgment B | Readiness, gaps and condition owners:
Actions and review dates:
Human start decision:
Actual downstream adoption:

Acceptance and takeover

Keep the judgments separate. Unknown never means passed. If quality is acceptable but conditions are missing, do not invent upstream rework. Resources cannot repair an unacceptable deliverable. Escalate conflicting criteria, insufficient evidence or actions outside authority.

Copyable task instruction

Compare the supplied deliverable with its acceptance criteria and readiness checklist. Produce separate quality and readiness judgments, citing evidence and marking unknowns. Assign quality corrections and readiness gaps to the appropriate owners. Do not approve a start or send materials externally. Use the output format above.

Repeatability check

Have another person check the same version independently. Compare both judgments and issue ownership. Investigate unclear standards and inputs when results differ; do not lower the criteria for a different executor.

Related research

← AI Skill overview