Skip to content

Latest commit

 

History

History
67 lines (49 loc) · 2.04 KB

File metadata and controls

67 lines (49 loc) · 2.04 KB

EXPLAINME: 007 Agent Meta-Language

A language where autonomous agents are first-class citizens

007 is a programming language for agents that make agents. Not a framework. Not a library. A language where autonomous agents are first-class citizens, agent creation is type-checked, and agent injection is a parse error.

— README.adoc

007 is built on six pillars: Harvard architecture (grammatical data/control separation), linear handles (agent leak prevention), session types (protocol verification), choreographic compilation (deadlock freedom), capability-based security, and content-addressable behaviour. See Language Spec and Design Rationale for the full treatment.

Caveat: This is a research language under active development. The compiler, interpreter, and semantic analyser each have mk2 rewrites in progress. Choreographic compilation is the highest-risk component. The dual-use classification is precautionary, not regulatory.

Path Proves

spec/grammar.ebnf

Grammar separates data_expr and control_stmt — agent injection is a parse error

spec/LANGUAGE-SPEC.adoc

Full language specification including type system

spec/OPERATIONAL-SEMANTICS.adoc

Operational semantics are formally defined

spec/TYPE-SYSTEM-SPEC.adoc

Type system spec covering linear handles and session types

compiler-core/

Rust compiler implementation exists

interpreter/

Interpreter with Harvard architecture enforcement

lexer/

Lexer for the 007 grammar

stdlib/

Standard library with agent primitives

tests/conformance/

Conformance tests validating language behaviour

examples/

Worked examples of agent definitions

DUAL-USE-NOTICE.adoc

Dual-use classification is documented and reasoned

docs/design-introspection-harvard.adoc

Harvard architecture design rationale with JTV lineage