product-apps-decomposition

Product decomposition divides a single product into multiple focused applications instead of a monolithic build to define app boundaries, mitigate single-point-of-failure reliability risks, and reduce LLM context bloat for more efficient token loading.

LLM context bloat happens when a monolithic build overloads the context window. Product decomposition solves this by splitting the product into focused apps, ensuring token-efficient LLM context loading and allowing independent feature deployability without exceeding limits.

The best way to validate a pre-specified product app breakdown is to use existing plans from docs/<app>/product.md, which automatically guides the decomposition to ensure app boundaries meet reliability and token economy requirements instead of generating generic guidance.

You need product decomposition during product planning when deciding whether to build a single monolithic application or multiple focused apps, specifically to mitigate LLM context bloat, reduce single-point-of-failure reliability risks, and define structured app boundaries.

Yes, you can use product decomposition without a pre-defined breakdown because it provides generic decomposition guidance focused on core reliability and token economy principles for LLM-powered workflows, helping you define app boundaries from scratch.