Mojentic.Agents.IterativeProblemSolver (Mojentic v1.2.0)

Copy Markdown View Source

An agent that iteratively attempts to solve a problem using available tools.

This solver uses a chat-based approach to break down and solve complex problems. It will continue attempting to solve the problem until it either succeeds, fails explicitly, or reaches the maximum number of iterations.

The solver uses the ChatSession to maintain conversation state and automatically handles tool calls through the broker. It monitors the LLM's responses for completion indicators ("DONE" or "FAIL") to determine when to stop iterating.

Usage

alias Mojentic.LLM.Broker
alias Mojentic.LLM.Gateways.Ollama
alias Mojentic.LLM.Tools.{DateResolver, AskUser}
alias Mojentic.Agents.IterativeProblemSolver

broker = Broker.new("qwen3:32b", Ollama)

solver = IterativeProblemSolver.new(
  broker,
  tools: [DateResolver, AskUser],
  max_iterations: 5
)

case IterativeProblemSolver.solve(solver, "What's the date next Friday?") do
  {:ok, result} -> IO.puts("Result: #{result}")
  {:error, reason} -> IO.puts("Error: #{inspect(reason)}")
end

Options

  • :tools - List of tool modules available to the LLM (default: [])
  • :max_iterations - Maximum number of iterations before giving up (default: 3)
  • :system_prompt - Custom system prompt (default: problem-solving assistant prompt)
  • :temperature - LLM temperature for response generation (default: 1.0)

Completion Indicators

The solver monitors the LLM's responses for these keywords:

  • "DONE" (case-insensitive) - Task completed successfully
  • "FAIL" (case-insensitive) - Task cannot be completed

When either indicator is detected, the solver requests a final summary and returns that to the caller.

Summary

Types

t()

Functions

new(broker, opts \\ [])

Creates a new IterativeProblemSolver.

solve(solver, problem)

Executes the problem-solving process.

Types

t()

@type t() :: %Mojentic.Agents.IterativeProblemSolver{
  broker: Mojentic.LLM.Broker.t(),
  max_iterations: pos_integer(),
  system_prompt: String.t(),
  temperature: float(),
  tools: [module()]
}

Functions

new(broker, opts \\ [])

@spec new(
  Mojentic.LLM.Broker.t(),
  keyword()
) :: t()

Creates a new IterativeProblemSolver.

Parameters

  • broker - The LLM broker to use for generating responses
  • opts - Keyword list of options:
    • :tools - List of tool modules (default: [])
    • :max_iterations - Maximum iterations (default: 3)
    • :system_prompt - Custom system prompt (default: problem-solving prompt)
    • :temperature - LLM temperature (default: 1.0)

Examples

broker = Broker.new("qwen3:32b", Ollama)

# With defaults
solver = IterativeProblemSolver.new(broker)

# With custom options
solver = IterativeProblemSolver.new(broker,
  tools: [MyTool],
  max_iterations: 5,
  system_prompt: "You are a specialized assistant.",
  temperature: 0.7
)

solve(solver, problem)

@spec solve(t(), String.t()) :: {:ok, String.t()} | {:error, term()}

Executes the problem-solving process.

This method runs the iterative problem-solving process, continuing until one of these conditions is met:

  • The task is completed successfully ("DONE")
  • The task fails explicitly ("FAIL")
  • The maximum number of iterations is reached

After the loop completes, the solver requests a summary of the final result, excluding process details.

Parameters

  • solver - The IterativeProblemSolver instance
  • problem - The problem or request to be solved

Returns

  • {:ok, summary} - Success with final result summary
  • {:error, reason} - Error from broker or chat session

Examples

{:ok, result} = IterativeProblemSolver.solve(solver, "Calculate the area of a circle with radius 5")
# => {:ok, "The area is approximately 78.54 square units."}

{:ok, result} = IterativeProblemSolver.solve(solver, "What's the weather tomorrow?")
# Uses tools iteratively to gather info and answer