Gouvernance des disciplines

VérifiéPrudence

Transformer une pratique récurrente entre capacités en une discipline Arcanum gouvernée : carte de discipline, entrée de catalogue et parcours de durcissement.

Spar Skills Guide Bot
DocumentationAvancé
23/07/2026
Claude Code
#governance#catalog#schema-validation#lifecycle#knowledge-management

Recommandé pour

Synthèse automatique

Transforme une pratique transversale récurrente en discipline Arcanum gouvernée : une fiche adossée à des preuves, une entrée de catalogue et un itinéraire de durcissement nommé, sans s'arroger l'autorité des sigils ou des sorts.

Points forts

  • Cycle de vie explicite en six modes (scan, formalize, route, validate, promote, deprecate) pilotés par un argument-hint unique.
  • Sépare nettement l'autorité « catalogue » des surfaces de promotion : le sigil ne crée ni ne modifie sigils, sorts, registres ou ontologies, il délègue au propriétaire.
  • Garde-fous anti-dérive intégrés : tag public-safe / private-parent avec catalogue cible, preuve externe obligatoire pour un motif actif, validateur de catalogue et verdict pass/flag/block.
  • Mode --dry-run et règle de preuve nommée avant toute montée de statut, ce qui rend les promotions auditables.

Limites

  • Très couplé à l'écosystème Arcanum (catalogue, schéma de fiche, validateur) : difficilement réutilisable hors de ce cadre.
  • Ne produit pas le durcissement lui-même ; toute demande de constitution, validateur ou sort renvoie vers un autre propriétaire.
  • Sans preuves existantes hors du sigil d'origine, une fiche reste bloquée en candidate et le mode promote ne peut pas aboutir.
Quand l'utiliser

À utiliser quand une même pratique se répète dans plusieurs sigils, sorts ou documents et n'a encore ni fiche ni entrée de catalogue.

Quand l'éviter

À éviter pour une tâche ponctuelle, une note locale à un seul sigil, ou une demande de modification directe d'un sigil, d'un sort ou d'un contrat de registre.

Analyse de sécurité

Prudence
Score qualité72/100

The skill is a documentation/governance workflow: it scans a repo for recurring practices, writes a discipline card and a catalog row, and routes hardening decisions to other owners. Nothing in it requests secret exfiltration, obfuscated payloads, curl|sh, destructive deletes, or disabling of safety. The residual risk is ordinary for a file-editing agent: Bash, Write, and Agent are powerful tools used here for plausible, legitimate bookkeeping in a docs directory. It also includes a --dry-run affordance, which is a positive signal. Hence 'caution' rather than 'safe' or 'risky'.

Points d'attention
  • •Declares Bash and Write, so it can modify files in the repository (discipline cards, catalog rows, development artifacts)
  • •Declares Agent, allowing it to spawn sub-agents whose behavior is not described in this file
  • •No explicit command allowlist, path allowlist, or prohibition on network access for the Bash calls
Comment cette analyse est faite

Exemples

Scan for hidden practices
/discipline-governance scan evidence-review — find practices that recur across the sigils and spells dealing with evidence review, propose candidate disciplines, and tag each one public-safe or private-parent with its target catalog.
Formalize a practice into a card
/discipline-governance formalize receipt-id-legend --evidence docs/receipts.md --dry-run — turn this practice into a discipline card plus catalog row at status candidate, and show me the proposed files before writing.
Validate then plan a promotion
/discipline-governance validate active-pattern evidence-review — check the card against the discipline schema and run the catalog validator; if it passes, tell me what the promote route would need to name (owner, evidence, validation surface, mutation boundary).

surface_kind: generated-native-runtime-package runtime: claude canonical_source: arcana/discipline-governance/SKILL.md alias_of: null generated_by: tools/bootstrap_arcanum.sh --profile mutation_policy: regenerate-from-canonical-source name: discipline-governance description: "Use when: formalizing, routing, validating, promoting, or retiring an Arcanum discipline so a recurring cross-capability practice gets a card, catalog entry, and a hardening route without claiming sigil or spell authority." argument-hint: "<scan|formalize|route|validate|promote|deprecate> <practice-or-discipline-id> [--status <status>] [--evidence <path>] [--dry-run]" tier: arcana domain: discipline-governance version: 0.2.1 origin: created from the disciplines layer (catalog, card template, schema, catalog validator) which had no owning lifecycle sigil; promoted to 0.2.0 on 2026-06-21 after the 3-regime experiment harness (low=gitignore, medium=receipt-id-legend, complex=scan+curation), telemetry emission, and a reflection pass; 0.2.1 applied the two reflection-named iterations (public/private candidate tagging in scan/route; active-pattern cross-sigil evidence rule in validate) allowed-tools: Read, Write, Glob, Grep, Bash, AskUserQuestion, Agent

Sigil: Discipline Governance

<objective> Turn a recurring cross-capability practice into a governed Arcanum discipline: an evidence-backed discipline card, a catalog entry, and a named hardening route, while keeping the disciplines layer a catalog authority rather than a promotion surface for sigils, spells, or knowledge. </objective> <logic-type> Arcana: long-lived governance of method-level practices across context selection, evidence, routing, validation, and promotion. </logic-type> <modes> - `scan`: find recurring or hidden practices that appear across multiple capabilities and propose candidate disciplines. **Each candidate is tagged `public-safe` or `private-parent` with a named target catalog** (public arcanum vs a private umbrella catalog); a practice tied to private composition must never be routed into the public catalog (v0.2.1). - `formalize`: turn one practice into a discipline card and a catalog row, default status `candidate`, written to the target catalog named by the candidate's `public-safe | private-parent` tag. - `route`: decide the hardening move for a discipline (catalog-only, template, validator, constitution, spell, or sigil) and hand off to the owning lifecycle. The route never moves a `private-parent` discipline into the public catalog (v0.2.1). - `validate`: run the catalog validator and check a card against the discipline schema, returning pass, flag, or block. **An `active-pattern` (or higher) discipline must cite at least one evidence ref OUTSIDE its own owning sigil; a card whose evidence is only its own sigil README is a thin one-per-sigil pointer and stays `candidate`** (v0.2.1). - `promote`: raise a discipline's status when the next route names owner, evidence, validation surface, and mutation boundary. - `deprecate`: retire a discipline with rationale and a superseding route. </modes> <applicability> Use this sigil when:
  • a practice recurs across multiple sigils, spells, or framework documents and has no formal home,
  • scattered rules for one practice are causing drift, rework, or confusion,
  • a discipline card or catalog entry must be created, corrected, or kept schema-valid,
  • a discipline needs a hardening decision (constitution, validator, template, spell, or sigil),
  • a discipline's status should advance or retire with named evidence,
  • the discipline catalog must be validated after edits. </applicability>
<non-applicability> Do not use this sigil when:
  • the practice is a one-off task, not a durable cross-capability method,
  • a capability-local note inside one sigil or spell is enough,
  • the request is to mutate a sigil, spell, registry, ontology, or glossary contract (route to that owner),
  • the request is to author the enforceable constitution itself (route to constitution-governance after the discipline names that route),
  • the request is to define a canonical term (route to definitions-governance). </non-applicability>
<inputs> Expected inputs, if available:
  • a candidate practice name or an existing discipline_id,
  • evidence paths where the practice already appears,
  • the current catalog disciplines/DISCIPLINES.md,
  • the card template disciplines/templates/discipline-card.md,
  • the schema disciplines/discipline.schema.yml,
  • prior scan output under disciplines/development/,
  • a desired status or hardening route. </inputs>
<chain-boundary> Discipline Governance owns the disciplines layer: cards, the catalog, scan evidence, routing decisions, and status changes.
  • constitution-governance owns constitutions. When a discipline's hardening route is a constitution, this sigil names the route and hands off; it does not author the constitution itself.
  • definitions-governance owns canonical terms a discipline depends on.
  • inventory and context-builder supply source-backed evidence; their evidence stays non-authoritative until an owner promotes it.
  • decision-gate resolves blocker-level promotion, precedence, or scope decisions.
  • sigil-development owns this sigil's own lifecycle, observability, and reflection.

A discipline may recommend a route, but it must never directly promote registry, ontology, glossary, sigil, or spell knowledge. </chain-boundary>

<default-output> Prefer target-local outputs:
  1. disciplines/cards/<discipline-id>.md for a discipline card,
  2. a single row in disciplines/DISCIPLINES.md for the catalog entry,
  3. disciplines/development/ for scan evidence and routing notes,
  4. arcana/discipline-governance/development/ for this sigil's own development artifacts.

Do not author constitutions, validators, or sigils inline; name the route and hand off to the owning lifecycle. </default-output>

<process> 1. Resolve the target: a candidate practice (for `scan`/`formalize`) or an existing `discipline_id` (for `route`/`validate`/`promote`/`deprecate`). 2. Classify the mode. If no mode is given, infer the smallest mode that satisfies the request and state the inference. 3. Gather evidence. A discipline needs at least one concrete repository reference. Confirm the maintenance signal: the practice appears across multiple capabilities, already has scattered rules, or has caused drift, rework, or confusion. Tag the candidate `public-safe` or `private-parent` and name its target catalog (v0.2.1). 4. For `formalize`, write a card from `disciplines/templates/discipline-card.md` with: status, steward, purpose, boundary, evidence refs, quality bar, and promotion guardrail. Add one catalog row to the target catalog (public `disciplines/DISCIPLINES.md`, or the private umbrella catalog for a `private-parent` discipline) matching the required columns. Default status is `candidate`. 5. For `route`, choose the smallest sufficient hardening move and name its owner: - catalog-only when the card is enough, - template when the practice needs a reusable shape, - validator when the practice can be checked deterministically, - constitution when the practice enforces structure or form (hand off to `constitution-governance`), - spell or sigil only when the practice needs executable lifecycle behavior. 6. For `validate`, run the catalog validator and check the card against the schema. Report pass, flag, or block with the failing rule. Additionally flag any `active-pattern`-or-higher discipline whose evidence is only its own owning sigil (a thin one-per-sigil pointer): it must cite cross-capability evidence outside that sigil or be downgraded to `candidate` (v0.2.1). 7. For `promote`, apply the Growth Rule: raise status only when the next route names owner, evidence, validation surface, and mutation boundary. Route blocker-level decisions through `decision-gate`. 8. For `deprecate`, mark the discipline `deprecated`, cite the rationale, and name the superseding route. 9. Preserve the layer boundary in every mode: do not mutate capability-local contracts and do not treat discipline evidence as canonical knowledge. 10. Validate the result (catalog validator, schema, local Markdown links, product-neutral wording) and return paths changed, status, route, and next step. </process> <discipline-model> A discipline has a named purpose, a boundary, evidence, a steward, a maturity status, and a next hardening move.

Status ladder:

  • candidate: useful practice exists, authority and validation still being proven,
  • active-pattern: already used by active capabilities, but the discipline-level contract is not canonical,
  • implemented: working repository support exists, may still need discipline-level rules,
  • canonical: accepted framework authority with validator or constitution support,
  • deprecated: superseded or withdrawn.

Growth Rule: promote a discipline only when the next route names its owner, evidence, validation surface, and mutation boundary. </discipline-model>

<routing-model> Map each discipline to the smallest sufficient hardening route:

| Route | Use when | Owner | | --- | --- | --- | | catalog-only | the card captures the practice and no enforcement is needed yet | Discipline Governance | | template | the practice needs a reusable shape | the template's host capability | | validator | the practice can be checked deterministically | tools/ plus the rule's owner | | constitution | the practice enforces structure or form across artifacts | constitution-governance | | spell or sigil | the practice needs executable lifecycle behavior | spellcraft or sigil-development |

A discipline can hold more than one route over time, but each route names a separate owner. </routing-model>

<validation-model> Every discipline card declares one validation mode, matching the schema:
  • prose-review: human or model review only,
  • validator: a deterministic script or check enforces it,
  • fixture: example-backed checks enforce it,
  • observability: usage signals confirm it,
  • mixed: a combination.

Run python3 disciplines/scripts/validate-discipline-catalog.py after any catalog edit. A card that claims a validator mode must cite an existing validator or be blocked until one exists. </validation-model>

<quality-bar> A successful execution must:
  • produce or update an evidence-backed discipline card and a schema-valid catalog row,
  • cite at least one concrete repository reference for the practice,
  • name the steward and the next hardening move,
  • choose the smallest sufficient route and hand off enforcement to the owning lifecycle,
  • keep discipline guidance separate from capability-local authority,
  • pass the catalog validator and resolve all local Markdown links,
  • route blocker-level promotion or scope decisions through decision-gate,
  • return paths changed, status, route, and next step. </quality-bar>
<anti-patterns> Avoid:
  • cataloging a one-off task as a durable discipline,
  • formalizing a practice with no concrete evidence,
  • letting a discipline promote a sigil, spell, registry, ontology, or glossary entry,
  • authoring a constitution, validator, or sigil inline instead of naming the route,
  • claiming a validator mode with no validator behind it,
  • raising status without owner, evidence, validation surface, and mutation boundary,
  • editing the catalog without running the catalog validator. </anti-patterns>
<observability> For meaningful executions, emit or prepare telemetry with:
  • mode,
  • target discipline id,
  • card created or updated,
  • catalog row added or changed,
  • route chosen and owner,
  • status before and after,
  • validator result,
  • decision gates required,
  • pass, flag, or block result. </observability>
<output-contract> Return:
## Discipline Governance Result

- Mode: <scan | formalize | route | validate | promote | deprecate>
- Target discipline: <id or candidate>
- Status: pass | flag | block
- Card: <path or none>
- Catalog row: <added | changed | none>
- Route: <catalog-only | template | validator | constitution | spell | sigil>
- Status change: <none or before -> after>
- Validator result: <pass | flag | block | not-run>
- Decisions needed: <none or list>
- Validation: <checks and result>
- Next route: constitution-governance | decision-gate | sigil-development | task-session | deferred
</output-contract> <origin> Created from the existing `disciplines/` layer (catalog, card template, schema, and catalog validator), which had no owning lifecycle sigil. Generalized into an Arcana governance sigil that formalizes, routes, validates, promotes, and retires disciplines without claiming sigil or spell authority. </origin> </content> </invoke>

Du fichier au premier essai

Le téléchargement contient exactement le contenu affiché sur cette fiche. Cette copie peut différer de la dernière version du dépôt source.

Consulter la source et ses fichiers annexes

Ce téléchargement contient uniquement SKILL.md. Les scripts, références, modèles et dépendances éventuels ne sont pas inclus. Récupérez les fichiers nécessaires depuis la source en conservant leur arborescence avant de poursuivre.

1. Télécharger le fichier

Télécharger SKILL.md

Depuis la racine d’un projet de test, enregistrez le fichier sous .skill-downloads/discipline-governance-mrx3xxv4/SKILL.md, ou utilisez la commande suivante. Ce dossier sert à examiner la skill avant son installation.

Commande pour macOS/Linux ou WSL : un shell POSIX et curl sont nécessaires. Sous Windows sans WSL, utilisez le bouton de téléchargement et créez le dossier manuellement.

mkdir -p '.skill-downloads/discipline-governance-mrx3xxv4' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/discipline-governance-mrx3xxv4/download' --output '.skill-downloads/discipline-governance-mrx3xxv4/SKILL.md.part' && mv '.skill-downloads/discipline-governance-mrx3xxv4/SKILL.md.part' '.skill-downloads/discipline-governance-mrx3xxv4/SKILL.md'

2. Vérifier les prérequis

Lisez SKILL.md avant de l’utiliser. Si le fichier cite des scripts ou services externes, examinez-les et préparez les accès nécessaires dans votre environnement de test.

Outils déclarés par l’auteur (à vérifier dans votre agent) : Read, Write, Glob, Grep, Bash, AskUserQuestion, Agent

Aucune section de prérequis reconnue dans ce fichier. Cela ne signifie pas qu’il est autonome : les bibliothèques, outils et accès peuvent être cités ailleurs dans le contenu ou dans le dépôt source.

Les plateformes de la fiche sont des indications du catalogue. Les chemins ci-dessous suivent la documentation des outils ; ils ne prouvent pas que cette skill y a été exécutée. Aucune version minimale de ces outils n’est attestée pour cette fiche.

3. Charger la skill

Après vérification, copiez le dossier complet dans le projet avec l’une des commandes ci-dessous, puis ouvrez une nouvelle session de l’outil. La commande s’arrête si un dossier du même nom existe déjà : comparez les fichiers avant de le remplacer.

Claude Code
test -f '.skill-downloads/discipline-governance-mrx3xxv4/SKILL.md' && test -s '.skill-downloads/discipline-governance-mrx3xxv4/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/discipline-governance' && cp -R '.skill-downloads/discipline-governance-mrx3xxv4' '.claude/skills/discipline-governance'

OpenSkills

OpenSkills doit déjà être installé. La procédure locale install/read est contrôlée avec OpenSkills 1.5.0, qui exige Node.js ≥ 20.6.0. Aucun slug du catalogue n’est résolu automatiquement.

Installation officielle d’OpenSkills
openskills install './.skill-downloads/discipline-governance-mrx3xxv4'

openskills read 'discipline-governance-mrx3xxv4'

La lecture doit afficher le contenu du fichier installé et son dossier de base. Elle ne lance ni la skill ni ses dépendances.

Pour un agent qui lit AGENTS.md, synchronisez ensuite l’index. Cette commande modifie AGENTS.md : relisez le diff avant de l’utiliser.

openskills sync

Mode manuel ou autre outil : joignez SKILL.md à une conversation, ou ouvrez-le dans un agent capable de lire les fichiers. Ajoutez les ressources annexes nécessaires. Il n’existe pas de dossier universel reconnu par tous les outils.

4. Faire un premier essai

Utilisez une entrée courte et fictive : un endpoint d’API, un petit document ou un brief d’interface. Dans une nouvelle conversation, joignez le fichier ou donnez son chemin à l’agent, puis adaptez cette demande :

Lis .skill-downloads/discipline-governance-mrx3xxv4/SKILL.md et applique ses instructions à cet exemple : [décrire une entrée de test]. Indique les fichiers, dépendances ou accès qui manquent. Produis le résultat demandé par la skill sans inventer de données ni exécuter de script non examiné.

Résultat attendu : une sortie correspondant aux instructions de la skill, ou une liste précise des prérequis manquants. Une simple confirmation de lecture ne valide pas le premier usage. Si une ressource externe est indisponible, ce parcours reste à tester.

Skills similaires