Ref-006: VibesPro™ Integration & Usage Guide

Overview

This guide explains how to use the integrated VibesPro™ system within the SEA™ monorepo to scaffold services, docs, and APIs in a governed way.

1. Quick Start

Generate a Service

1
2

# Non-interactive generation using defaults
just sea:gen service --name orders --domain commerce --language python

This runs:

  1. Copier (Generates files)
  2. Semantic Validator (Checks CADSL)
  3. CALM Validator (Checks Policies)
  4. KG Projector (Updates Graph)

Run Local LLM Stack

SEA™ uses a local-first AI stack.

1
2
3
4
5

# Start the local model server (Ollama)
just llm:start

# Start the Gateway (LiteLLM with Policy)
just llm:proxy

2. Configuration (answers.sea.yml)

The generation is controlled by a standardized answer file. Key sections:

AI & Workflows

1
2

include_ai_workflows: true
enable_temporal_learning: true

Tech Stack

1
2
3

architecture_style: "hexagonal"
backend_framework: "fastapi"
app_framework: "next"

3. Type Synchronization

We enforce a Python -> TypeScript flow for shared types.

  1. Define models in apps/<service>/domain/models.py (Pydantic).
  2. Run sync: 
    1

    just sea:type-sync
  3. Consume in apps/web/src/types/ (Auto-generated TS interfaces).

4. Governance Gates

Your PR will be blocked by sea-governed-gen if: