apresmoi/up-extract-standards

Extract coding standards from an example repository

Scan an example repository and extract its patterns into a standards file for the standards library.

Input

$ARGUMENTS should be the path to an example repository to scan. If empty, ask the user.

Procedure

1. Scan the repository:

   • Detect language(s) and framework(s)
   • Read package.json, tsconfig.json, pyproject.toml, .eslintrc*, .prettierrc*, ruff.toml, and similar config files
   • Examine folder structure (depth 3 levels)
   • Sample 5-10 representative source files to detect patterns

2. Extract patterns across these categories:

   Project structure:

   • Root folder layout
   • Source code organization (by feature, by type, hybrid)
   • Where tests live relative to source
   • Config file locations

   Naming conventions:

   • File naming (camelCase, kebab-case, PascalCase)
   • Folder naming
   • Component/class/function naming
   • Export naming (named vs default)

   File patterns:

   • Typical file length range
   • One-component-per-file or grouped
   • Barrel exports (index.ts re-exports)
   • Co-located tests, styles, types

   Import/export patterns:

   • Absolute vs relative imports
   • Path aliases (@/ etc.)
   • Named exports vs default exports
   • Import ordering conventions

   Tooling:

   • Linter + config (ESLint, Ruff, etc.)
   • Formatter (Prettier, Black, etc.)
   • Type checker (TypeScript strict mode, mypy, etc.)
   • Test runner + patterns (Jest, Vitest, pytest, etc.)
   • Build tool

   Library-specific patterns:

   • Styling approach (Tailwind, CSS Modules, styled-components, etc.)
   • State management
   • API/data fetching patterns
   • Middleware patterns (if backend)

   Anti-patterns observed:

   • Note any inconsistencies or patterns that seem intentionally avoided

3. Determine the right standards file to write or update:

   • Match to existing files in .unpack/standards/ based on detected stack
   • If no match, propose a new file name

4. Write the standards file with status seeded and all extracted patterns.

5. Report what was extracted, confidence level on each pattern, and ask the user to confirm or adjust key patterns before marking as confirmed.

Important

• This is a read-only scan — do NOT modify the example repo
• Distinguish between patterns the repo uses consistently vs one-off occurrences
• If the repo is inconsistent, note the inconsistency rather than picking a side