L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

pymc-bayesian-modeling

pymc-bayesian-modeling

Compatibility alias for the descriptive PyMC skill name. Delegate to the canonical local `pymc` payload while preserving route and README compatibility.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: pymc-bayesian-modeling description: Compatibility alias for the descriptive PyMC skill name. Delegate to the canonical local pymc payload while preserving route and README compatibility.

pymc-bayesian-modeling (Compatibility Alias)

Purpose

Provide a stable descriptive alias for Bayesian modeling workflows that are canonically maintained under the sibling pymc skill directory.

This preserves:

  • README-facing descriptive naming
  • existing skills-lock and catalog entries
  • route compatibility for callers that still ask for pymc-bayesian-modeling

Resolution Order

  1. Use the canonical local pymc skill payload first.
  2. Reuse its assets, references, and scripts:
    • ../pymc/SKILL.md
    • ../pymc/assets/**
    • ../pymc/references/**
    • ../pymc/scripts/**
  3. Keep this alias thin and free of its own duplicated heavy payload.

Minimal Workflow

  1. Read ../pymc/SKILL.md for the full PyMC workflow.
  2. Use the canonical templates and diagnostic scripts from ../pymc/.
  3. Report results under the requested alias name only for user-facing continuity.
Skills enregistrées