# Create Rule Processor

> Scaffold a new rule processor, implement the evaluation logic, and write tests.

- Skill: `majiayu000/create-rule-processor` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/create-rule-processor`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/create-rule-processor/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/create-rule-processor

---


1. **Scaffold the rule** using the built-in script:

   ```bash
   pnpm create-rule <name> <scope> <description>
   ```

   - `<name>`: kebab-case rule name (e.g., `vehicle-validation`)
   - `<scope>`: processor group (`mass-id`, `credit-order`, or `mass-id-certificate`)
   - `<description>`: short description of what the rule validates

   This creates the standard file structure:

   ```
   {rule-name}/
   ├── {rule-name}.processor.ts
   ├── {rule-name}.lambda.ts
   ├── {rule-name}.processor.spec.ts
   ├── {rule-name}.lambda.e2e.spec.ts
   ├── {rule-name}.test-cases.ts
   ├── index.ts
   ├── project.json
   └── jest.config.ts
   ```

2. **Implement the processor**: In `{rule-name}.processor.ts`, implement the `evaluateResult()` method. The processor extends `ParentDocumentRuleProcessor<RuleSubject>`.

3. **Write test cases**: In `{rule-name}.test-cases.ts`, define shared test data using `@faker-js/faker` and `zocker`. Never use real data (no real company names, tax IDs, addresses, etc.).

4. **Write unit tests**: In `{rule-name}.processor.spec.ts`, test the core evaluation logic using the shared test cases.

5. **Write e2e tests**: In `{rule-name}.lambda.e2e.spec.ts`, test the Lambda handler end-to-end.

6. **Apply to a methodology** (if needed):

   ```bash
   pnpm apply-methodology-rule <methodology> <rule> <scope>
   ```

   Example: `pnpm apply-methodology-rule carbon-organic vehicle-validation mass-id`

7. **Verify**: Run `pnpm lint:affected && pnpm ts:affected && pnpm test:affected` to confirm everything passes.