ASTRA Syntax at a Glance

This page is a quick map of the main forms in an ASTRA agent. Follow the links for explanations and examples.

Agent structure

An agent is declared in a .astra file whose name matches the agent class. Its body can contain module declarations, type declarations, initial beliefs and goals, plans, inference rules, and teleo-reactive functions:

agent Hello {
    module Console console;

    plan +!main(list args) {
        console.println("Hello");
    }
}

The keyword rule can be used in place of plan; the two are equivalent. The ASTRA archetype and the LLM libraries use rule.

See AgentSpeak(L) and ASTRA for the agent model and Declaring Variables for variable forms.

Common plan triggers

Form	Meaning
`+belief(...)`	A belief was added.
`-belief(...)`	A belief was removed.
`+!goal(...)`	A goal was adopted.
`^!goal(...)`	An adopted goal failed.
`+?formula(...)`	A test goal needs a plan to find an answer.
`@message(performative, sender, content)`	A message arrived.

Plans may include a context after : and a body in braces. See Beliefs and Inference, Encapsulated Goals, Test Goals, Failure Handling, and Agent Communication.

Types and expressions

Declare formula signatures in a types block. Common types include int, long, float, double, char, boolean, string, list, and funct. See Types and Operators for the full table, expression precedence, and examples.

Statements

Common statements include sub-goals (!goal(...)), test goals (?formula(...)), belief updates (+belief(...) and -belief(...)), queries (query(...)), conditionals (if (...) { ... }), loops (while (...) { ... }), and module actions (module.action(...)). See Variables, Choice and Repetition for control flow and Failure Handling for try/recover.

The statements activate function(...); and deactivate; start and stop a teleo-reactive function. See Teleo-Reactive Functions.