oris-runtime 0.61.0

An agentic workflow runtime and programmable AI execution system in Rust: stateful graphs, agents, tools, and multi-step execution.
use thiserror::Error;

use crate::{language_models::LLMError, output_parsers::OutputParserError, prompt::PromptError};

#[derive(Error, Debug)]
pub enum ChainError {
    #[error("LLM error: {0}")]
    LLMError(#[from] LLMError),

    #[error("Retriever error: {0}")]
    RetrieverError(String),

    #[error("OutputParser error: {0}")]
    OutputParser(#[from] OutputParserError),

    #[error("Prompt error: {0}")]
    PromptError(#[from] PromptError),

    #[error("Missing Object On Builder: {0}")]
    MissingObject(String),

    #[error("Missing input variable: {0}")]
    MissingInputVariable(String),

    #[error("Serde json error: {0}")]
    SerdeJsonError(#[from] serde_json::Error),

    #[error("Incorrect input variable: expected type {expected_type}, {source}")]
    IncorrectInputVariable {
        source: serde_json::Error,
        expected_type: String,
    },

    #[error("Error: {0}")]
    OtherError(String),

    #[error("Database error: {0}")]
    DatabaseError(String),

    #[error("Agent error: {0}")]
    AgentError(String),

    /// Human-in-the-loop: execution paused for approval; payload has action_requests and review_configs.
    #[error("Interrupt (human-in-the-loop)")]
    Interrupt(serde_json::Value),
}