Documentation Search MCP Server

Instructions

Suggest libraries based on partial input for auto-completion.

Args:
    partial_name: Partial library name to search for (e.g. "lang" -> ["langchain"])

Returns:
    List of matching library names

Implementation Reference

• src/documentation_search_enhanced/main.py:692-724 (handler)

  The handler function for the 'suggest_libraries' MCP tool. It provides auto-completion suggestions for library names by matching against the configured docs_urls dictionary, prioritizing exact matches, then prefix matches, then substring matches, returning up to 10 sorted suggestions.

  async def suggest_libraries(partial_name: str):
      """
      Suggest libraries based on partial input for auto-completion.
  
      Args:
          partial_name: Partial library name to search for (e.g. "lang" -> ["langchain"])
  
      Returns:
          List of matching library names
      """
      if not partial_name:
          return list(sorted(docs_urls.keys()))
  
      partial_lower = partial_name.lower()
      suggestions = []
  
      # Exact matches first
      for lib in docs_urls.keys():
          if lib.lower() == partial_lower:
              suggestions.append(lib)
  
      # Starts with matches
      for lib in docs_urls.keys():
          if lib.lower().startswith(partial_lower) and lib not in suggestions:
              suggestions.append(lib)
  
      # Contains matches
      for lib in docs_urls.keys():
          if partial_lower in lib.lower() and lib not in suggestions:
              suggestions.append(lib)
  
      return sorted(suggestions[:10])  # Limit to top 10 suggestions

• src/documentation_search_enhanced/main.py:692-692 (registration)

  The @mcp.tool() decorator registers the suggest_libraries function as an MCP tool.

  async def suggest_libraries(partial_name: str):