build

SpecForge Build Orchestrator

Apply spec changes, run code generation, implement handlers, test and iterate until everything works.

Overview

The build process is orchestrated in phases:

1. Apply Spec Changes - Update OpenAPI and database schemas
2. Code Generation - Generate type-safe code from schemas
3. Handler Implementation - Implement business logic in parallel
4. Test & Iterate - Run tests, observe behavior, fix issues
5. Integration - Wire everything together and verify

Phase 1: Apply Spec Changes

Apply Database Migrations

Read the SpecForge configuration from CLAUDE.md to identify the database plugin:

# Extract database plugin from CLAUDE.md
grep "database:" CLAUDE.md | cut -d: -f2 | tr -d ' '

Use the database plugin's migration skill:

Invoke the {database-plugin}/migrations-expert skill with:
- action: "apply-migrations"
- migrations_dir: "migrations/"
- database_url: from environment or docker-compose

Steps:

1. List pending migrations
2. Backup database (if production)
3. Apply migrations in order
4. Verify schema matches expected state
5. Handle errors (rollback if needed)

Verify OpenAPI Spec

Use the openapi-expert skill to validate the spec:

Invoke openapi-expert skill with:
- action: "validate-spec"
- spec_path: "spec/openapi.yaml"

Validation checks:

• Valid OpenAPI 3.x syntax
• All $ref references resolve
• Schemas are well-formed
• Examples match schemas
• No duplicate operation IDs

Phase 2: Code Generation

Run Codegen Pipeline (DB Schema → Types)

Identify the codegen plugin from CLAUDE.md:

grep "codegen:" CLAUDE.md | cut -d: -f2 | tr -d ' '

Invoke the codegen plugin's generation skill:

Use {codegen-plugin}/codegen-expert skill with:
- action: "generate-from-schema"
- schema_dir: "migrations/"
- output_dir: "{backend-path}/src/generated/db"
- database_url: from environment

This generates:

• Type-safe database models (structs, classes, types)
• Query functions with compile-time verification
• Database client code
• Schema types and enums