AgentSkillsCN

refactor-lexical-ontology

审计标识符和命名空间,确保词法语义和本体论正确性。 检测语义角色错位(代理人/工具/过程/制品)、派生误用 (例如,-er代理名词),以及模块、包、类、函数和数据制品之间的类别错误。输出一份包含可操作修复建议的Markdown报告。

SKILL.md
--- frontmatter
name: refactor-lexical-ontology
license: MIT
description: |
  Audit identifiers and namespaces for lexical-semantic and ontological correctness.
  Detect semantic role misalignment (agent/tool/process/artifact), derivational misuse
  (e.g., -er agent nouns), and category errors between modules, packages, classes,
  functions, and data artifacts. Output a single Markdown report with actionable fixes.
metadata:
  author: Jordan Godau
  references:
    - 01_INTENT.md
    - 02_PRECONDITONS.md
    - 03_RULES.md
    - 04_PROCEDURE.md
    - 05_OUTPUT.md
  keywords:
    - naming
    - ontology
    - lexical
    - semantic
    - agent
    - process
    - artifact
    - suffix
    - convention

Instructions

Read all references in references/ before using this skill.

Signals

  • The user says: "naming feels off", "planner vs planning", "module naming convention", "agentive suffix misuse", "semantic role mismatch", "ontology drift", "lexical audit"
  • A naming refactor is planned or ongoing
  • Confusion exists between thing vs doer vs process vs tool in identifiers
  • The codebase has inconsistent conventions between directories/modules/classes/functions

References

Directory: references/

  • 01_INTENT.md
  • 02_PRECONDITIONS.md
  • 03_RULES.md
  • 04_PROCEDURE.md
  • 05_OUTPUT.md

Scripts

If scripts are present, also read scripts/INDEX.md and follow any automated steps first.