Skip to content

Concepts

High-level ideas that guide agex’s design and usage.

  • The Big Picture - Agents that think in code, curated micro‑DSLs, and sandbox philosophy.
  • Security - Walled‑garden security model, AST validation, safe introspection.
  • Nearly Python - The Python subset available to agents — what works, what’s different.
  • A Note on MCP - Why agex focuses on code‑native integration vs. tool schemas.

Use this section to understand why agex is built the way it is, and which constraints enable safe, powerful runtime interoperability.