Implementation:Run llama Llama index Download Pack

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

Overview

Provides functions for downloading LlamaPack templates from the llama_index packs repository, installing their dependencies, and dynamically loading them into memory.

Description

The download/pack.py module handles the retrieval of LlamaPack templates from the llama-index-packs directory within the main llama_index GitHub repository. Unlike the general module downloader, this module is specifically designed for the newer pack template format that uses pyproject.toml for dependency management.

The download workflow consists of two key functions:

1. download_module_and_reqs takes the local and remote directory paths along with a package name and sub-module name. It recursively discovers all source files under the pack's namespace (llama_index/packs/{sub_module}) using get_source_files_recursive, downloads each file, and writes them to the local directory. It then downloads the pack's pyproject.toml and creates a minimal README.md (required for pip install from pyproject.toml). Finally, it installs the pack's dependencies by running pip install . in the pack directory using a ChangeDirectory context manager.
2. download_llama_pack_template is the primary entry point. It converts a package name (e.g., "llama-index-packs-resume-screener") into a Python sub-module name by stripping the prefix and replacing hyphens with underscores. It calls download_module_and_reqs and then dynamically loads the module using importlib to return the requested pack class.

The module also provides track_download for sending analytics events to the LlamaHub proxy server.

Usage

Use this module when you need to download a LlamaPack template for customization or direct use. This is the backend for the download_llama_pack convenience function and the llamaindex-cli download-llamapack CLI command. It is particularly useful when you want a complete, editable copy of a pack template rather than just installing it as a dependency.

Code Reference

Source Location

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

Signature

def download_llama_pack_template(
    new_install_parent: str,
    llama_pack_class: str,
    llama_pack_url: str = LLAMA_PACKS_CONTENTS_URL,
    llama_pack_source_files_dir_path: str = LLAMA_PACKS_SOURCE_FILES_GITHUB_TREE_URL,
    refresh_cache: bool = False,
    custom_dir: Optional[str] = None,
    custom_path: Optional[str] = None,
    base_file_name: str = "__init__.py",
) -> Any

Import

from llama_index.core.download.pack import download_llama_pack_template

I/O Contract

Inputs

Name	Type	Required	Description
new_install_parent	str	Yes	The pip package name of the pack (e.g., "llama-index-packs-resume-screener").
llama_pack_class	str	Yes	The Python class name of the pack to load (e.g., "ResumeScreenerPack").
llama_pack_url	str	No	Base URL for pack raw content on GitHub. Defaults to the llama-index-packs directory.
llama_pack_source_files_dir_path	str	No	GitHub tree URL for enumerating source files. Defaults to the llama_index main branch tree 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.
base_file_name	str	No	Entry point file name for the pack. Defaults to "__init__.py".

Outputs

Name	Type	Description
result	Any	The downloaded LlamaPack class, ready to be instantiated.

Usage Examples

from llama_index.core.download.pack import download_llama_pack_template

# Download a LlamaPack template
ResumeScreenerPack = download_llama_pack_template(
    new_install_parent="llama-index-packs-resume-screener",
    llama_pack_class="ResumeScreenerPack",
)

# Instantiate and use the pack
pack = ResumeScreenerPack(resume_path="resume.pdf")
response = pack.run(query="What are the candidate's skills?")

Related Pages

• Environment:Run_llama_Llama_index_Python_LlamaIndex_Core