This is just a little bot I made, honestly just wanted alt text to describe images when I cant open them in public (for moderation), but I figured while I was plumbing in gemini I might as well let you chat with it and discuss the images too.
- It will scan every url found, and respond to the ones that are supported images
- messages starting with its displayname it will respond to.
- Messages starting with the image generation model (
hidream
on this instance) will generate an image - use
{displayname}, forget
to clear context window - use
{displayname}, help
to get back here
{
"gemini-api": "[REDACTED]"
API token for gemini, as per the gemini api docs
"nanogpt-api": "[REDACTED]",
API token for NanoGPT integration
"nanogpt-image-model": "hidream",
The model name to use for NanoGPT image generation
"nanogpt-image-w": 1360,
"nanogpt-image-h": 768,
Width and height settings for generated images (in pixels)
"max_file_len": 10
Maximum file size (in megabytes) of the images to download and send to the gemini api
"jid": "gemini@pain.agency",
JID (string) with localpart and server for the bot
"password": "[REDACTED]",
"displayname": "gemini",
Password and Displayname for the bot's login
"rooms": [
"testing@group.pain.agency",
"chaos@group.pain.agency",
]
}
List of rooms/mucs to join on startup