llama-index-tools-videogen

Official LlamaIndex tools for the VideoGen API. Give any LlamaIndex agent the ability to turn a script into a finished narrated video, plus generate standalone images, video clips, voiceovers, sound effects, music, and talking-head avatars.

Every tool wraps the official videogen SDK and runs as the team that owns your API key.

Install

pip install llama-index-tools-videogen

Setup

Get an API key from app.videogen.io/api and set it as VIDEOGEN_API_KEY:

export VIDEOGEN_API_KEY="your_api_key"

Usage

from llama_index.core.agent.workflow import FunctionAgent
from llama_index.llms.openai import OpenAI
from llama_index.tools.videogen import create_videogen_tools

agent = FunctionAgent(
    tools=create_videogen_tools(),
    llm=OpenAI(model="gpt-4o"),
    system_prompt="You produce short narrated videos with VideoGen when asked.",
)

response = await agent.run(
    "Make a short narrated video explaining why staying hydrated matters."
)
print(response)

By default each tool blocks until the workflow or tool execution finishes and returns the completed result.

Options

from videogen import VideoGen
from llama_index.tools.videogen import create_videogen_tools, VideoGenToolSpec

tools = create_videogen_tools(
    # Provide a pre-configured client, or pass api_key / base_url, or rely on env.
    client=VideoGen(api_key="your_api_key"),

    # Return the run/execution id immediately instead of waiting.
    wait=False,

    # Only expose a subset of tools to the model.
    include=["script_to_video", "generate_image", "text_to_speech"],
)

# Or use the tool spec:
tools = VideoGenToolSpec().to_tool_list()

Tools

• Workflows: cancel_workflow_run, get_workflow_run, list_workflow_runs, prompt_to_video_clip, script_to_video, slideshow_to_video, storyboard_to_video, video_to_video, voiceover_to_video.
• Media generation: cancel_tool_execution, generate_avatar, generate_image, generate_motion_graphic, generate_music, generate_sound_effect, generate_video_clip, get_tool_execution, image_3d_effect, list_tool_executions, remove_image_background, remove_video_background, text_to_speech, upscale_image, upscale_video, vectorize_image.
• Projects: export_project, get_project, list_project_remix_actions, list_projects, remix_project.
• Files: get_file, list_files, upload_file.
• Resources and account: get_me, list_languages, list_tts_voices.

Avatar generation and supported video workflows use reusable actor entities. Pass actor_entity_id and set avatar_quality to LOW, STANDARD, HIGH, or MAX. Narrator changes are passed in remix_actions as CHANGE_NARRATOR action dictionaries.

Documentation

• Use with AI agents
• MCP server
• API reference

License

MIT

Account and tool permissions

Generation uses your VideoGen account credits. Obtain user approval before invoking generation, editing or export tools. Start with a restricted include list when an agent only needs read access. File upload tools send the selected file to VideoGen; keep API keys in server-side environment variables and never place them in shared agent transcripts.

For a credential-free smoke check, build the tools with a synthetic key without invoking them. The package tests mock or avoid external calls; they do not prove paid generation.

VideoGen · MCP setup

2.5.4

Fix polling after workflow, media-tool and export starts to use the Python SDK’s normalized response keys. No API capabilities or dependency requirements changed.