Code Generator

Before generating a single line, inspects your repo — stack, framework, naming conventions, import style, error handling patterns. Every generated file fits in as if a team member wrote it, not an AI that guessed.

Example tasks

• →Read our repo and add a user notification system matching our existing service pattern
• →Inspect our codebase and generate API client types from our OpenAPI spec
• →Audit our code style then add a new route following the same structure

Implements a complete feature end-to-end in one pass: database migration, backend service and route, frontend component consuming the API, and integration tests. Opens a single PR with every layer linked.

Example tasks

• →Add a team invitations feature — Postgres migration, REST endpoint, React invite form, and tests
• →Build a file upload flow: S3 presigned URLs endpoint, upload progress component, and error handling
• →Add real-time notifications: WebSocket server route, frontend hook, and toast component

Builds a complete new microservice from zero: package.json, TypeScript config, health check endpoint, structured logging, env validation, Dockerfile, and GitHub Actions CI. Everything runnable on day one.

Example tasks

• →Scaffold a payments microservice: Fastify + Prisma + Stripe webhooks + Docker + CI pipeline
• →Create a new email-sender service with queue processing and retry logic
• →Build an analytics ingestion service: endpoint, validation, ClickHouse writer, and README

Reads your existing test setup — framework, mocking patterns, fixture style — then generates meaningful tests for untested code. Unit tests for pure functions, integration tests for routes, and edge cases you'd miss in a code review.

Example tasks

• →Generate a full Vitest test suite for our services directory with 80%+ coverage
• →Write integration tests for all our Express routes using our existing supertest setup
• →Add edge case tests for our billing calculation module — boundary values and error paths

Migrates your codebase from one technology to another without breaking the public API. Reads every affected file before starting, migrates in dependency order, and cleans up the old package when done.

Example tasks

• →Migrate our JavaScript Express app to TypeScript — strict mode, no any
• →Convert our Mongoose models to Prisma — same schema, same relation names
• →Move from CommonJS require() to ESM imports across the whole backend

Takes an architecture document, ERD, or OpenAPI spec and generates the full implementation — every file, every layer. No placeholders, no stubs. If a decision is ambiguous, it makes a principal-level choice and documents the assumption.

Example tasks

• →Generate a complete REST API from our architecture doc — all 12 endpoints with validation and tests
• →Build the backend from our OpenAPI spec: types, route handlers, middleware, and DB layer
• →Implement all entities from our ERD as Prisma models with proper indexes and relations

Takes an entity-relationship diagram or data model and generates typed ORM models, SQL migration files, seed data, and query helpers. Uses the ORM already in your repo — Prisma, Drizzle, TypeORM, SQLAlchemy, or ActiveRecord.

Example tasks

• →Generate Prisma schema and migration for our 8-entity ERD
• →Create Drizzle table definitions and relations from our data model diagram
• →Add a new "workspaces" entity to our existing Prisma schema with proper cascades