FoundationAgents/MetaGPT · error · NotImplementedError

Invalid strategy: {strategy}, only support BFS/DFS/MCTS curr

Error message

Invalid strategy: {strategy}, only support BFS/DFS/MCTS currently!

What it means

TreeofThought._initialize_solver maps the Strategy enum to solver classes: BFS->BFSSolver, DFS->DFSSolver, MCTS->MCTSSolver. Any other strategy value hits an explicit NotImplementedError. In practice this fires when a strategy is injected that is not one of the three enum members (note MCTSSolver.solve itself is also unimplemented upstream).

Source

Thrown at metagpt/strategy/tot.py:264

    def _initialize_solver(self, strategy):
        """
        Initialize the solver based on the chosen strategy.

        Args:
            strategy (Strategy): The strategy to use for solving.

        Returns:
            ThoughtSolverBase: An instance of the appropriate solver.
        """
        if strategy == Strategy.BFS:
            self.solver = BFSSolver(config=self.config)
        elif strategy == Strategy.DFS:
            self.solver = DFSSolver(config=self.config)
        elif strategy == Strategy.MCTS:
            self.solver = MCTSSolver(config=self.config)
        else:
            raise NotImplementedError(f"Invalid strategy: {strategy}, only support BFS/DFS/MCTS currently!")

    async def solve(self, init_prompt=""):
        """
        Solve the problem using the specified strategy.

        Args:
            init_prompt (str): The initial prompt for the solver.
            strategy (str): The strategy to use for solving.

        Returns:
            Any: The solution obtained using the selected strategy.
        """
        await self.solver.solve(init_prompt)

View on GitHub (pinned to 11cdf466d0)

Solutions

1. Use Strategy.BFS or Strategy.DFS (these have working solvers)
2. Avoid Strategy.MCTS for now — MCTSSolver.solve raises NotImplementedError at runtime
3. To add a custom strategy, subclass TreeofThought and extend _initialize_solver with your solver class

Example fix

// before
tot = TreeofThought(strategy=Strategy.MCTS)  # constructs, but solve() raises NotImplementedError

// after
from metagpt.strategy.tot import TreeofThought, Strategy
tot = TreeofThought(strategy=Strategy.BFS)
answer = await tot.solve(prompt)

from metagpt.strategy.tot import Strategy
assert strategy in (Strategy.BFS, Strategy.DFS), "MCTS solver is not implemented; use BFS or DFS"

from metagpt.strategy.tot import Strategy

def is_implemented_strategy(strategy) -> bool:
    return strategy in (Strategy.BFS, Strategy.DFS)

try:
    tot = TreeofThought(config=cfg, strategy=strategy)
    answer = await tot.solve(prompt)
except NotImplementedError as e:
    if "only support BFS/DFS/MCTS" in str(e) or "solve" in str(e):
        tot = TreeofThought(config=cfg, strategy=Strategy.BFS)
        answer = await tot.solve(prompt)
    else:
        raise

When it happens

Trigger: Constructing TreeofThought(strategy=<something not Strategy.BFS/DFS/MCTS>) — e.g. a raw string that bypassed validation or a custom enum member added to Strategy without a solver.

Common situations: Passing strings like "beam" or "best-first" expecting support; extending the Strategy enum without registering a solver; note also that choosing MCTS currently raises NotImplementedError at solve() time since MCTSSolver.solve is a stub.

Related errors

• File format not supported.
• not supported action_type: {action_type}
• Invalid request type. Valid types: {', '.join([t.value for t
• Unsupported react mode: {self.rc.react_mode}

AI-assisted analysis of FoundationAgents/MetaGPT@11cdf466d0 (2026-08-14). Data as JSON: /api/errors/df5704a586086a6d. Report an issue: GitHub.