Context

As the technical scope of my projects grows, I need to manage multiple related concerns: UI code, infrastructure-as-code (Terraform), and shared scripts. Splitting these into separate repositories would introduce significant friction:

• Dependency Drift: Keeping shared tooling (like mise) and linting configurations in sync across multiple repos is painful.
• Complexity: It requires setting up and maintaining an artifact store (npm/pypi packages) to share code, which is overkill for a personal project where velocity is paramount.
• Context Switching: "Agentic" development works best when the AI has the full context of the system available in one place.

Decision

I decided to adopt a Monorepo structure.

This centralizes all code—frontend, infrastructure, and tooling—into a single repository.

Consequences

Pros

• Single Point of Reference: One place to look for everything. Documentation, code, and infrastructure live side-by-side.
• Shared Tooling & Standards: I can enforce common linting, formatting, and mise configurations at the root level.
• Reusable Workflows: CI/CD workflows and secrets can be defined once and reused across different projects within the repo.
• Velocity: Avoids the complexity of versioning and publishing internal packages. I can change a shared utility and update its consumers in a single atomic commit.
• Agentic Optimization: AI agents can inspect the entire codebase to understand cross-component dependencies and impact, leading to better code generation and debugging.

Cons

• Build Complexity: The build pipeline needs to be "smart" enough to only build/deploy what changed (handled via tools like Turbo or simple path filtering).
• Clone Size: The repo is larger than a single service repo, but negligible for this scale.