Implementation:Run llama Llama index Download Module

Knowledge Sources	Run_llama_Llama_index
Domains	Download, Module
Last Updated	2026-02-11 19:00 GMT

Overview

Provides the core functionality for downloading LlamaHub modules (loaders, tools, packs, and agents) from the remote repository, installing their dependencies, and dynamically loading them into memory.

Description

The download/module.py module is the primary mechanism for fetching community-contributed modules from LlamaHub. It defines a MODULE_TYPE enum with four categories: LOADER, TOOL, LLAMAPACK, and DATASETS.

The download workflow consists of three key functions:

1. get_module_info resolves a module class name to its module ID and list of extra files by consulting a library.json file. It first checks a local cache and falls back to fetching from the remote LlamaHub repository.
2. download_module_and_reqs downloads the module's base Python file and any extra files from the remote repository. If an __init__.py extra file is found, it automatically updates the local package's exports. It then reads a requirements.txt file and installs any missing dependencies using pip via subprocess.
3. download_llama_module is the primary entry point that orchestrates the full download flow: initializing a local directory, fetching module info, downloading files and installing requirements, and dynamically loading the module using importlib to return the requested class.

The module also provides track_download which sends an analytics event to the LlamaHub proxy server to track download counts for community modules.

The use_gpt_index_import parameter exists as a legacy compatibility flag for import rewriting, and skip_load allows downloading without loading the module into memory.

Usage

Use this module when you need to dynamically download and load community-contributed loaders, tools, or agent packs from the LlamaHub repository. It is the backend for the download_loader, download_tool, and similar convenience functions throughout the LlamaIndex framework.

Code Reference

Source Location

• Repository: Run_llama_Llama_index
• File: llama-index-core/llama_index/core/download/module.py

Signature

def download_llama_module(
    module_class: str,
    llama_hub_url: str = LLAMA_HUB_URL,
    refresh_cache: bool = False,
    custom_dir: Optional[str] = None,
    custom_path: Optional[str] = None,
    library_path: str = "library.json",
    base_file_name: str = "base.py",
    use_gpt_index_import: bool = False,
    disable_library_cache: bool = False,
    override_path: bool = False,
    skip_load: bool = False,
) -> Any

Import

from llama_index.core.download.module import download_llama_module

I/O Contract

Inputs

Name	Type	Required	Description
module_class	str	Yes	The name of the LlamaHub module class to download (e.g., "GmailOpenAIAgentPack").
llama_hub_url	str	No	Base URL for the LlamaHub remote repository. Defaults to the llama-hub main branch raw content URL.
refresh_cache	bool	No	If True, skips local cache and re-downloads from remote. Defaults to False.
custom_dir	Optional[str]	No	Custom directory name under the parent folder for downloads.
custom_path	Optional[str]	No	Custom absolute directory path for downloads.
library_path	str	No	Relative path to the library.json metadata file. Defaults to "library.json".
base_file_name	str	No	Name of the main Python file to download. Defaults to "base.py".
use_gpt_index_import	bool	No	Legacy flag for import rewriting compatibility. Defaults to False.
disable_library_cache	bool	No	If True, does not write library.json to local cache. Defaults to False.
override_path	bool	No	If True, writes files directly to the base directory instead of a module_id subdirectory. Defaults to False.
skip_load	bool	No	If True, downloads the module but does not load it into memory. Returns None. Defaults to False.

Outputs

Name	Type	Description
result	Any	The downloaded module class (e.g., a Loader, Pack, or Agent class), or None if skip_load is True.

Usage Examples

from llama_index.core.download.module import download_llama_module

# Download and load a module from LlamaHub
WikipediaReader = download_llama_module(
    module_class="WikipediaReader",
    refresh_cache=False,
)

# Use the loaded module
reader = WikipediaReader()
documents = reader.load_data(pages=["Python (programming language)"])

Related Pages

• Environment:Run_llama_Llama_index_Python_LlamaIndex_Core