Crate simple_agent_type 

Core types and traits for SimpleAgents.

This crate provides all foundational types, traits, and error definitions shared across the SimpleAgents project. It is a pure types crate with no runtime, HTTP, or I/O dependencies.

Architecture

SimpleAgents follows a trait-based architecture:

• Provider: Trait for LLM provider implementations
• Cache: Trait for caching responses
• RoutingStrategy: Trait for provider selection

Main Types

• [Message]: Role-based conversation messages
• [CompletionRequest]: Unified request format
• [CompletionResponse]: Unified response format
• [ApiKey]: Secure API key handling

Example

use simple_agent_type::prelude::*;

// Create a request
let request = CompletionRequest::builder()
    .model("gpt-4")
    .message(Message::user("Hello!"))
    .temperature(0.7)
    .build()
    .unwrap();

// Access request properties
assert_eq!(request.model, "gpt-4");
assert_eq!(request.messages.len(), 1);

Features

• Type Safety: Strong types prevent common errors
• Transparency: All transformations tracked via [CoercionFlag]
• Security: API keys never logged ([ApiKey])
• Validation: Early validation with clear errors
• Async: All traits use async_trait

Re-exports

pub use error::HealingError;

pub use error::ProviderError;

pub use error::Result;

pub use error::SimpleAgentsError;

pub use error::ValidationError;

Modules

cache

Cache trait for response caching.

coercion

Coercion types for tracking response healing.

config

Configuration types for SimpleAgents.

error

Error types for SimpleAgents.

message

Message types for LLM interactions.

prelude

Prelude module for convenient imports.

provider

Provider trait and types.

request

Request types for LLM completions.

response

Response types for LLM completions.

router

Routing strategy trait for provider selection.

tool

Tool calling types for function/tool integrations.

validation

Validation types for sensitive data.