pub fn parse_react_step(response: &str) -> Result<ReActStep, AgentError>Expand description
Parses a raw LLM response string into a ReActStep.
Supported formats:
Thought: <text>Action: <tool_name>(<input>)Action: <tool_name>: <input>Final Answer: <text>
ยงErrors
Returns AgentError::ParseError if the response does not match any known prefix.