ai-storyteller 0.1

The AI Storyteller - where your imagination meets state-of-the-art AI creativity.

AI Storyteller - where your imagination meets state-of-the-art AI creativity.

MCP

Locally

We need two instances:

• MCP server

• MEDIA server

ATM, the Marimo app has MEDIA server functionality. For local inference, we can use that.

As a pre-requisite, both processed, need to be given this:

export DISABLE_HTTPS_ENFORCEMENT=1
export DISABLE_AUTHENTICATION=1
export OPENAI_API_KEY='ollama'
export OPENAI_BASE_URL='http://127.0.0.1:11434/v1'

export STORYTELLER_MEDIA_PATH='/Users/me/repos/ai-storyteller/media'
export STORYTELLER_MEDIA_URL='http://localhost:8000/media/'

export STORYTELLER_OUTPUT_PATH='/Users/me/repos/ai-storyteller/output'
export STORYTELLER_OUTPUT_MEDIA_URL='http://localhost:8000/media/output/'

export STORYTELLER_TOOL_MODEL_PROVIDER='openai'
export STORYTELLER_TOOL_MODEL_NAME='gpt-5-mini'
export STORYTELLER_STORY_MODEL_NAME='o4-mini'
export STORYTELLER_IMAGE_DESCRIPTOR_MODEL_NAME='gpt-4.1-mini'

Then start the MEDIA server (FastAPI/Playground):

make run

Then start the MCP server in HTTP mode (FastAPI):

make mcp

To start the AI chat (with storyteller tool), type:

make chat

License

MIT

Authors

• Artur Barseghyan

• Dale Richardson