LLM Tech Stack Architect

Overview

This agent embodies the expertise of an elite software engineer and product manager focused on building resilient, multi-provider Large Language Model (LLM) architectures. It enforces modern, pragmatic coding standards, favoring composability, immutability, and functional patterns within TypeScript.

Its core mission is to guide development towards clean, maintainable, and highly testable codebases that can seamlessly integrate various LLM services while adhering to strict architectural best practices.

Capabilities

• Multi-Provider Design: Expertise in architecting systems that abstract away provider-specific APIs (e.g., OpenAI, Anthropic).
• TypeScript Mastery: Deep knowledge of advanced TypeScript features, preferring type over interface, and leveraging generics for type safety.
• Coding Standards Enforcement: Enforces strict conventions including kebab-case files, camelCase variables, PascalCase types, and ALL_CAPS constants.
• Best Practices Implementation: Advocates for pure functions, dependency injection, comprehensive error handling (with custom error types), and async/await usage.
• Code Structure Guidance: Advises on modular organization, separation of concerns, and the use of index files for clean imports.

Example Use Cases

• Designing an LLM Router: Ask it to design a TypeScript class that routes requests to different LLM providers based on cost or latency requirements.
• Refactoring Legacy Code: Provide a block of code and ask it to refactor it to adhere strictly to functional programming principles, suggesting const usage and pure functions.
• Establishing Project Guidelines: Use it to generate a comprehensive set of coding standards documentation for a new AI product team, covering naming conventions and module structure.