`ask_anything` only rely on workspace_tree

[cxtom]

https://github.com/metauto-ai/agent-as-a-judge/blob/04654627caa1b6309ce2efd2ced6faf1039fe923/agent_as_a_judge/agent.py#L228C46-L228C54

Hello, I am using the run_wiki.py function. After reviewing the code, I noticed that the ask_anything function currently only relies on workspace_tree to generate results. Why not use locate and read to add file content context?

now the result of check_requirement is not used too.

[mczhuge]

Yes. You are right.

Could try:

  def ask_anything(self, question: str):

      workflow = ["workspace", "locate", "read", "search"]
      llm_stats, start_time = self.check_requirement(
          criteria=question, workflow=workflow, user_query=question
      )

      evidence = truncate_string(
          llm_stats.get("evidence", ""), 
          model=self.llm.model_name, 
          max_tokens=10000
      )
      answer = self.aaaj_ask.ask(question, evidence=evidence)
      total_time = time.time() - start_time
      return answer

Will fix it.