intent-engine

Intent classification for Claude Code extensions works by walking a decision tree to resolve the extension type, event details, and context from natural language descriptions. It outputs a structured intent-resolution payload for concierge orchestration.

Converting developer workflow descriptions into structured extension requests involves an intent engine that detects compound needs and resolves type, event, and context. This produces a complete payload to hand off rich context to the generator pipeline.

Automatic and on-demand intent classification for Claude Code hooks resolves event details and characteristics from user descriptions. The engine maps the classified intent to the correct extension type using a behavioral decision tree.

No, you cannot use the intent-engine directly to generate extensions. It is a knowledge skill loaded by the extension-concierge and is not directly invocable by users, serving instead to prepare the concierge payload for the generator pipeline.

No, you do not need to provide structured input for Claude Code intent classification. You can provide a natural-language developer workflow description and let the Intent Engine classify and prepare the concierge payload automatically.

The limitation of using a decision tree for intent classification is that it operates within predefined automatic and on-demand contexts to resolve type, event, and context. It functions as a knowledge skill and cannot be directly invoked by users for standalone queries.