Jido.AI.Reasoning.AlgorithmOfThoughts.Strategy (Jido AI v2.2.0)

Copy Markdown View Source

Algorithm-of-Thoughts (AoT) execution strategy for Jido agents.

This strategy performs one-pass algorithmic reasoning with in-context search exemplars and extracts a structured AoT result.

Summary

Functions

Returns the last AoT result from strategy state.

Returns the action atom for handling LLM partial deltas.

Returns the action atom for handling LLM results.

Returns the action atom for handling request lifecycle rejections.

Returns the action atom for starting AoT exploration.

Types

config()

@type config() :: %{
  model: String.t(),
  profile: profile(),
  search_style: search_style(),
  temperature: float(),
  max_tokens: pos_integer(),
  examples: [String.t()],
  require_explicit_answer: boolean(),
  llm_timeout_ms: pos_integer() | nil
}

profile()

@type profile() :: :short | :standard | :long

search_style()

@type search_style() :: :dfs | :bfs

Functions

get_result(agent)

@spec get_result(Jido.Agent.t()) :: map() | nil

Returns the last AoT result from strategy state.

llm_partial_action()

@spec llm_partial_action() :: :aot_llm_partial

Returns the action atom for handling LLM partial deltas.

llm_result_action()

@spec llm_result_action() :: :aot_llm_result

Returns the action atom for handling LLM results.

request_error_action()

@spec request_error_action() :: :aot_request_error

Returns the action atom for handling request lifecycle rejections.

start_action()

@spec start_action() :: :aot_start

Returns the action atom for starting AoT exploration.