Generated from smart-base/skills/content/authoring-who-smart-guidelines/l3-readiness.md — do not edit here.
✎ Edit this page’s source

l3-readiness

Skill id: l3-readiness · Package: authoring-who-smart-guidelines · The check before Gateway_L3Ready of smart-guideline-end-to-end, task Task_L3Ready of smart-guideline-end-to-end.bpmn. Run by the authoring agent; it records no verdict on any asset.

This is the order rule (owner, 2026-10-15: “order strict, sign-off QA”) as a check with its own contract. It is not a stage: it produces no asset, only the answer the gateway takes. The owner gave it a skill of its own on 2026-10-16, so that the check has named inputs and outputs like every stage around it.

Input and output

The contract is input.schema.json and output.schema.json.

In: the l1Kg (stage 2), the l2Assets (every stage-4 l2-dak-authoring pass’s artifacts), the reviews requested so far, one per extraction or L2 asset, and l3Planned, the L3 artefacts stage 6 is about to write, each with its sourceModel.
Out: ready, which is the gateway’s outcome; findings, each naming the asset, the problem and the stage that fixes it; checked, every asset looked at, so that no findings is a determined empty and not a skipped check.

Steps

L1. Every source the use case includes has an l1Kg extraction, at library/<bib>/l1-extraction/. A source without one is no-l1-extraction, fixed by stage 2.
L2 on L1. Every L2 asset names at least one L1 node in wasDerivedFrom. One that names none is no-l1-upstream, fixed by stage 4.
L3 on L2. Every l3Planned entry’s sourceModel is an l2Assets path. One that is not is no-l2-upstream, fixed by stage 4. Source documents (a schedule poster, a slide deck) are stage-2 input, never a sourceModel.
Review requested. Every extraction and every L2 asset has a reviews entry: its review item opened and staged. One without is review-not-requested, fixed by the stage that made it. The verdict is not awaited: adjudication is QA, not a gate.
Answer. ready is true exactly when findings is empty. The gateway then takes «yes» to stage 6, or «no» back to stage 4 (which fills a missing extraction from stage 2).

A green build and passing tests do not make a finding go away. An L3 asset with no L2 upstream is a finding, not a shortcut.

Processes that run this skill

process	step(s) that name it
SMART Guideline end to end	Check L3 readiness