MCP-Repo2LLM

Instructions

Implementation Reference

• mcp-repo2llm-server.py:15-32 (handler)

  The tool handler function for 'get_gitlab_repo', decorated with @mcp.tool() for registration, which creates a GitlabRepo2Txt instance and invokes its process_repo method to retrieve and format the repository contents as text.

  @mcp.tool()
  async def get_gitlab_repo(repo_url: str, branch: str = "master")->str:
      """
      Process and return the code from a GitLab repository branch as text
      """
      try:
          repo_processor = GitlabRepo2Txt()
          repo_name, content = repo_processor.process_repo(
          repo_url=repo_url,
          branch=branch  # optional parameter
          )
          # logger.info(f"Processed GitLab repository: {repo_name}")
          # logger.info(f"Processed GitLab content: {content}")
          return content
      except Exception as e:
          # print(f"Error processing GitLab repository: {e}")
          return None

• repo2llm/gitlibrepo2txt.py:116-151 (helper)

  The process_repo method of the GitlabRepo2Txt class, which performs the core work of fetching README, repository structure, and non-binary file contents from a GitLab repository using the GitLab API.

  def process_repo(self, repo_url, branch='master'):
      """
      处理GitLab仓库并返回处理后的内容
      
      Args:
          repo_url (str): GitLab仓库URL
          branch (str, optional): 分支名称. 默认为 'master'
          
      Returns:
          tuple: (repo_name, content_string) - 仓库名和处理后的内容字符串
      """
      repo_name = repo_url.split('/')[-1]
      repo = self.gitlab.projects.get(repo_url.replace('https://gitlab.com/', ''))
  
      # print(f"Getting README for {repo_name}")
      readme_content = self._get_readme_content(repo, branch)
  
      # print(f"\nGetting repository structure for {repo_name}")
      repo_structure = f"Repository structure: {repo_name}\n"
      repo_structure += self._traverse_repo_iteratively(repo)
  
      # print(f"\nGetting file contents for {repo_name}")
      file_contents = self._get_file_contents_iteratively(repo, branch)
  
      instructions = "Use the following files and contents for analysis:\n\n"
      
      # 组合所有内容
      content = (
          instructions +
          f"README:\n{readme_content}\n\n" +
          repo_structure +
          '\n\n' +
          file_contents
      )
      
      return repo_name, content