vibe-coding-yaml
AI outputs YAML descriptors instead of code, CLI executes deterministically for 100% auditable behavior with zero slop. Use when auditability is critical, generated code review needed, or deterministic execution required. YAML serves as intermediate representation. Triggers on "auditable generation", "YAML output", "deterministic execution", "code review", "vibe coding".
When & Why to Use This Skill
This Claude skill enables deterministic and 100% auditable code generation by utilizing YAML descriptors as an intermediate representation. It eliminates 'AI slop' by separating creative intent from execution, allowing developers to review structured YAML instead of raw code, ensuring high reliability and reproducibility in software engineering workflows.
Use Cases
- High-Compliance Development: Generate code in regulated industries where every change must be audited and verified through structured descriptors.
- Infrastructure as Code (IaC): Create deterministic cloud configurations and deployment scripts that are easy to version-control and peer-review.
- UI Component Libraries: Use YAML to define UI properties and styles, ensuring that CLI-generated components remain consistent with design systems.
- API Boilerplate & Routing: Streamline backend development by generating standardized API endpoints and validation logic from structured YAML definitions.
- Vibe Coding with Precision: Bridge the gap between rapid AI prototyping and production-grade engineering by adding a deterministic execution layer.
| name | vibe-coding-yaml |
|---|---|
| description | AI outputs YAML descriptors instead of code, CLI executes deterministically for 100% auditable behavior with zero slop. Use when auditability is critical, generated code review needed, or deterministic execution required. YAML serves as intermediate representation. Triggers on "auditable generation", "YAML output", "deterministic execution", "code review", "vibe coding". |
Vibe Coding YAML
Purpose
AI generates YAML descriptors instead of code, CLI executes deterministically for 100% auditable behavior with zero slop.
When to Use
- Auditable code generation
- Deterministic outputs required
- Generated code review needed
- UI component generation
- Infrastructure as code
- Compliance requirements
Core Instructions
Pattern
AI outputs YAML:
component:
type: Button
props:
text: "Submit"
variant: "primary"
onClick: "handleSubmit"
styles:
padding: "12px 24px"
borderRadius: "8px"
backgroundColor: "#007bff"
CLI executes deterministically:
generate-component component.yaml > Button.tsx
Benefits
- 100% auditable: Review YAML, not generated code
- Zero slop: Deterministic execution
- Versionable: YAML in Git
- Reproducible: Same YAML = same output
- Reviewable: Easier to review descriptors than code
Example Workflow
# 1. AI generates descriptor
descriptor = {
'type': 'api_endpoint',
'method': 'POST',
'path': '/users',
'handler': 'createUser',
'validation': {
'email': 'required|email',
'name': 'required|string'
}
}
# 2. Save as YAML
with open('endpoint.yaml', 'w') as f:
yaml.dump(descriptor, f)
# 3. Execute deterministically
$ generate-endpoint endpoint.yaml > routes/users.ts
# 4. Review YAML (not generated code)
$ git diff endpoint.yaml
Performance
- 100% auditable
- 0% slop rate
- Deterministic behavior
- Easy review process
Version
v1.0.0 (2025-10-23)