Manager Coach

Audience: T3 function and department management. They own procedures, resources and priorities inside a function. Their learning is mostly at Supervision (L3) and Organisational (L4) level.

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 leading-indicator guidance in output-style.md.

Gather

PMLC select <ws> --tier T3 --brief
PMLC select <ws> --tier T3 --level 3 4 --top-rated
PMLC select <ws> --relation owner absent --brief          # systems they own; functions that did not exist
PMLC interfaces <ws> --top 15

If the user names a function, add --entity <F-id>; otherwise produce one section per function with T3 findings, largest first.

For each function, write

• What your system allowed: three to six system learnings (procedures missing or not used, schedules that slid, records that hid information, resources short, roles absent).
• Interfaces you share: from interfaces, the seams with other functions and what to agree with them.
• Recommendations: what should change, control tier, evidence.
• Leading indicators: two to four, measurable from systems the function already has (CMMS, permits, training records, audit tracker).
• 90-day agenda: first 30 days (verify the condition still exists, quick containment), 60 (system change drafted with the people who do the work), 90 (indicator in the monthly report). No names or dates unless asked.

Output

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

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

Sections → kinds. Write one section per function (id = the function, e.g. mechanical-maintenance), with each part as a group:

• What your system allowed: learning ×3–6 (owner = the function that owns the change, not who "was involved"; see learning-principles.md).
• Interfaces you share: para.
• Recommendations: recommendation (change, hoc_tier, barrier, indicator).
• Leading indicators: indicator ×2–4.
• 90-day agenda: table (Days, Action; row_cites).

End with a summary section: one table with columns Function, Top learning, Top recommendation and Indicator.

Set "scope": {"tiers": ["T3"]}. Put data-quality corrections in data_quality, findings left out on purpose in exclusions, and same-condition findings in supporting_cites.

{"kind": "learning", "group": "What your system allowed", "lesson": "Return to service relied on memory, not a checklist.",
 "explanation": "…", "owner": "Maintenance function", "cites": ["FND-…"]}