ashley-eyekyamashley-eyekyam/probematrix-learnings-coach/skills/cluster-learning-dossier/SKILL.md

cluster-learning-dossier

Skill0 starsChanged 27 days agoMITView on GitHub

Build the complete learning dossier for one department or function (a Learning Cluster) from a ProbeMatrix investigation export - its practices, conditions, supervisory and organisational learnings, interfaces, barriers, warning signs, messages for each level from frontline to leadership, recommendations, training needs and transferability. Use when someone asks "what does maintenance (or operations, engineering, fire services, HSE, procurement…) need to learn from this investigation", wants a department-level lessons pack, or wants every learning for one function in one place.

In this repository24 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- auditor-assurance-packSkill
- barrier-learningsSkill
- board-briefSkill
- causal-mapSkill
- emergency-response-learningsSkill
- engineering-integrity-learningsSkill
- frontline-learning-cardsSkill
- interface-learningsSkill
- investigator-coachSkill
- learning-clustersSkill
- learnings-coachSkill
- learnings-registerSkill
- manager-coachSkill
- organisation-learning-mapSkill
- precursor-timelineSkill
- report-comparisonSkill
- safety-committee-guideSkill
- site-leadership-coachSkill
- supervisor-coachSkill
- supply-chain-contractor-learningsSkill
- top-management-briefSkill
- training-designerSkill

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Write the pack <run>/outputs/dossier-<F-id>.json (schema: learnings-coach/references/pack-schema.md.
line 46

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

Cluster Learning Dossier

Steps 2 and 3 of the method for one Learning Cluster: open it along sub-divisions S1–S14 (learnings-coach/references/ontology.md §4) and write one learning piece per sub-division.

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

Gather

PMLC select <ws> --entity <F-id> --out <ws>/sel-<F-id>.md              # full text
PMLC select <ws> --entity <F-id> --level 1 --brief                      # per level as needed
PMLC interfaces <ws>                                                     # for S6
PMLC timeline <ws> --before incident                              # for S8

For large clusters (more than ~60 findings) work level by level so each pass stays grounded; read the full justification and causal text of the highest-rated findings first.

Write one piece per sub-division

Piece	Grounded in	Content
S1 Profile	`classify` output	counts by level, top-rated count, tiers and relations involved
S2 Practice (L1)	L1 findings	the practices that contributed, why each made sense at the time
S3 Conditions (L2)	L2 findings	the conditions this entity created or tolerated
S4 Supervision (L3)	L3 findings	what supervisors should see, check, correct and refuse
S5 System (L4)	L4 findings	systems, resources, policies, norms that must change
S6 Interfaces	`interfaces.json` pairs involving this entity	what fell between this entity and others; who must talk to whom
S7 Barriers	all	barriers this entity owns, prevention or mitigation, how each failed
S8 Precursors	timeline candidates mentioning this entity's work	dated warnings and what was done with them
S9 Tier messages	all	one short message each for T1, T2, T3, T4 (and T5 if relevant)
S10 Recommendations	S2–S8	what should change, barrier, evidence, control tier, leading indicator
S11 Competence	S2–S5	competences to build, by tier
S12 Transferability	all	same site / other sites / industry-wide, with reason
S13 Comparison	only if a comparator report is supplied	counterpart yes/partial/none (use `report-comparison`)
S14 Formats	S9, S10	which learnings become Cards, Brief paragraphs, toolbox talks

Each piece: three to eight learnings, each one sentence of lesson plus two to four sentences of explanation, with the evidence trail. Merge findings that describe the same condition into one learning.

Output

Write the pack <run>/outputs/dossier-<F-id>.json (schema: learnings-coach/references/pack-schema.md; pack = dossier-<F-id>, e.g. dossier-F02 or dossier-F09b), then:

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

The bottom line holds the top five learnings for this entity. Give each sub-division its own section, with id = s<n>-<name>. The engine adds the evidence trail. Sections and the kinds that fill them:

S1: a table of counts.
S2–S5: learning ×3–8 each (lesson, explanation).
S6: para and learning.
S7: barrier (type, state, failure_mode, hoc_tier, fix).
S8: a timeline with signal, received_by and response.
S9: para, with group = the tier.
S10: recommendation.
S11: a table with columns Tier and Competence.
S12: a table with columns Learning, Reach and Reason.
S13: para, only when there is a comparator report.
S14: a table with columns Learning and Format.

Set "scope": {"entities": ["<F-id>"]}. Put data-quality corrections in data_quality, findings left out on purpose in exclusions, and same-condition findings in supporting_cites. For all clusters, write one dossier per cluster, largest first, plus a pack dossier-index with one table (Cluster, Findings, Top learning, File).

{"kind": "learning", "lesson": "Supervisors had no checklist to refuse a restart against.", "explanation": "…", "cites": ["FND-…"]}

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free