Training Designer

Audience: training and L&D. They turn learnings into competence that lasts. Training is the weakest control on its own: design it alongside the system fixes, not instead of them.

Engine: PMLC = python3 <skills-root>/learnings-coach/scripts/pmlc.py (this skill needs the learnings-coach skill installed alongside it). Read learnings-coach/references/learning-principles.md.

Gather

PMLC select <ws> --text "train|competen|qualif|knowledge|experience|familiar|skill" --brief
PMLC select <ws> --level 1 2 --brief            # acts and individual conditions

Read <ws>/clusters.json for tiers and entities (the hub's classify step builds it; never re-run classify here).

Write

1. Training-needs map: rows = role groups (entity × tier), columns = competences implicated by the findings; each cell cites findings. Mark which needs are real competence gaps and which are system problems that training cannot fix (send those to manager-coach).
2. Modules (three to six), each: audience, objectives (observable), duration, the case material used (de-identified, roles only), activities, and how competence is checked.
3. Scenario exercises: three decision-point scenarios from the event ("the meter is back from repair and the test rig is unavailable: what do you do?"), with facilitator notes.
4. Assessment: ten to fifteen questions with answer keys and the finding each tests; include questions that ask for the level of a cause (L1–L4).
5. Cascade plan: who trains whom, materials, and a measure of transfer at 90 days.

Output

Write the pack <run>/outputs/training-designer.json (schema: learnings-coach/references/pack-schema.md), then:

PMLC check-pack <ws> <run>/outputs/training-designer.json   # fix every problem it lists
PMLC pack <ws> <run>/outputs/training-designer.json         # -> training-designer.html (primary) and .md

Sections → kinds:

1. table with columns Role group, Competence and Need ("Competence gap" or "System problem: send to manager-coach"), plus row_cites.
2. module ×3–6 (title, objective, audience, duration, content, assessment).
3. scenario ×3, with the facilitator notes in prompts.
4. question ×10–15: why = what the question tests, answer = the answer key, good_evidence = what a good answer includes, cites = the finding it tests.
5. table with columns Who, Trains whom, Materials and Transfer measure at 90 days.

Set "scope": {"levels": [1, 2]}. Put data-quality corrections in data_quality, findings left out on purpose in exclusions, and same-condition findings in supporting_cites. If the user wants slides or a document, build them from this pack with the host's document tools.

{"kind": "module", "title": "Return to service", "objective": "…", "audience": "Fitters", "duration": "45 min",
 "content": ["…"], "assessment": ["…"], "cites": ["FND-…"]}