Three Extension Injection Points
- Entity ID:
ent-20260423-e005f100000b - Type:
concept - Scope:
private - Status:
active
Description
From the design guide: every agent loop has three places where extensions can intervene. (1) assemble() — what the model sees (instructions, tool schemas). (2) model() — what the model can reach (available tools). (3) execute() — whether/how an action runs (permission gates, pre/post hooks). A useful cross-framework lens for classifying extension mechanisms.
Key claims
- Every agent loop has three universal extension injection points: assemble, model, execute
Relations
- none yet