praxis-types

Core types and event model for the Praxis AI agent framework.

Overview

This crate provides the foundational types used across all Praxis crates:

• Event Model: StreamEvent enum for real-time streaming
• Configuration: Shared configuration types
• Message Types: Structured message representations
• Common Traits: Shared traits for extensibility

Features

• Zero-copy event streaming with StreamEvent
• Serde serialization support
• Chrono-based timestamps
• UUID-based identifiers

Installation

[dependencies]
praxis-types = "0.1"

Usage

StreamEvent

The core event type for streaming AI agent responses:

use praxis_types::StreamEvent;

match event {
    StreamEvent::Message { content, .. } => {
        println!("Message: {}", content);
    }
    StreamEvent::Reasoning { content, .. } => {
        println!("Reasoning: {}", content);
    }
    StreamEvent::ToolCall { name, arguments, .. } => {
        println!("Calling tool: {} with {}", name, arguments);
    }
    StreamEvent::ToolResult { result, .. } => {
        println!("Tool result: {}", result);
    }
    StreamEvent::Done { .. } => {
        println!("Stream complete");
    }
}

Configuration

use praxis_types::Config;

let config = Config {
    model: "gpt-4o".to_string(),
    temperature: 0.7,
    max_tokens: Some(1000),
};

Event Model

Praxis uses a canonical event model for streaming:

• Message: Assistant's response content (token-by-token)
• Reasoning: Internal reasoning/thinking process
• ToolCall: Function/tool invocation request
• ToolResult: Result from tool execution
• Done: Stream completion marker

This unified event model works across all components (LLM, Graph, API).

Part of Praxis Framework

This crate is part of the Praxis AI Agent Framework:

• praxis-graph - React agent orchestrator
• praxis-llm - LLM client (OpenAI, Azure)
• praxis-mcp - Model Context Protocol client
• praxis-persist - MongoDB persistence

License

MIT