Frontline Learning Cards

Audience: T1 frontline. They control how the task is done today. They do not control procedures, budgets or design, so never hand them an organisational fix as if it were theirs.

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 and the Learning Card and toolbox talk blocks in learnings-coach/references/output-style.md.

Gather

PMLC select <ws> --tier T1 --brief                     # what frontline roles were involved in
PMLC select <ws> --tier T1 --level 1 2                  # acts and conditions at the sharp end
PMLC select <ws> --tier T1 --top-rated           # the ones that mattered most

If the user names a crew or trade ("maintenance fitters", "control-room operators"), add --entity <F-id>.

Choose the learnings

Pick five to eight learnings that a frontline person can notice or act on during work: a warning sign, a hold point, a test that must not be skipped, a situation where they should stop and ask. Skip learnings that only managers can act on, but if one explains why the job felt normal (e.g. "this substitute part had been used for years"), mention it so people recognise the trap.

Write

One Learning Card per learning (title, what happened, why it made sense at the time, what the organisation is changing, warning signs, what good looks like, three discussion questions, evidence trail). Use the words crews use. Roles only, no names. Keep each card under 250 words.

Then two or three toolbox talks (five minutes each) that group the cards by task (e.g. "returning equipment to service", "first minutes of a gas release").

Close with a "stop and ask" list: the situations from this event where anyone on the crew should pause the job and ask a supervisor, with the support they should expect when they do.

Output

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

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

Sections → kinds:

• Learning Cards: card ×5–8 (title, what_happened, why_made_sense and organisation_changing (both required in this skill; check-pack warns when organisation_changing is missing in a T1 pack), warning_signs, good_looks_like, discussion ×3; must cite).
• Toolbox talks: one para per talk, with group = the talk's task, then a scenario for its discussion.
• "Stop and ask": check items, with when = the situation.

Set "scope": {"tiers": ["T1"]}. Put data-quality corrections in data_quality, findings left out on purpose in exclusions, and same-condition findings in supporting_cites.

{"kind": "card", "title": "Test what you disturb", "what_happened": "…", "why_made_sense": "…",
 "organisation_changing": "…", "warning_signs": ["…"], "good_looks_like": "…", "discussion": ["…", "…", "…"], "cites": ["FND-…"]}

Facilitator note (put it in a para before the cards). These cards are read by the people whose acts appear in the findings. Consult those involved before wide release. Never single out "the" operator of a small crew: if only one person held a role, write the task, not the post. Do not close a card with "be more careful" or "stay vigilant"; say what the organisation is changing instead (organisation_changing).

If translation is requested, translate the text fields and keep every ID unchanged.