Skip to main content

AgentScript

A small language for explicit, scoped, auditable LLM context.

Start TutorialsRead Blog
Traditional append-only chat versus AgentScript scoped context boundaries

Explicit context

Choose exactly what enters each model call with scoped use declarations.

Auditable runs

Trace model calls, tool outputs, memory access, and prompt context boundaries.

Structured agents

Build multi-step workflows with typed inputs, contracts, tools, memory, and agents.