DescribedTypes.jl

Annotate Julia types/functions and generate JSON Schemas for LLM provider APIs.

Overview

DescribedTypes.jl lets you attach human-readable descriptions to Julia struct types and Julia function methods, then automatically produces JSON Schema dictionaries compatible with LLM structured-output APIs.

For functions, the package also supports argument execution from JSON payloads: function -> schema and JSON -> typed function call.

Supported adapters:

Installation

using Pkg
Pkg.add("DescribedTypes")

Or in the Pkg REPL:

pkg> add DescribedTypes

Quick Start

using DescribedTypes
using JSON

struct Person
    name::String
    age::Int
end

DescribedTypes.annotate(::Type{Person}) = Annotation(
    name="Person",
    description="A person.",
    parameters=Dict(
        :name => Annotation(name="name", description="The person's name", enum=["Alice", "Bob"]),
        :age  => Annotation(name="age", description="The person's age"),
    ),
)

# OpenAI structured-output format
schema_dict = schema(Person, llm_adapter=OPENAI)
print(JSON.json(schema_dict, 2))

{
  "name": "Person",
  "description": "A person.",
  "strict": true,
  "schema": {
    "type": "object",
    "properties": {
      "name": {
        "type": "string",
        "description": "The person's name",
        "enum": [
          "Alice",
          "Bob"
        ]
      },
      "age": {
        "type": "integer",
        "description": "The person's age"
      }
    },
    "required": [
      "name",
      "age"
    ],
    "additionalProperties": false
  }
}

Function tools quickstart:

function ping(city::String; verbose::Bool=false)
    return (; city, verbose)
end

tool_schema = schema(ping, llm_adapter=OPENAI_TOOLS)
call_result = callfunction(ping, Dict("city" => "Paris"))
(tool_schema["name"], call_result)

("ping", (city = "Paris", verbose = false))

See the Guide for more detailed usage and examples.