Filters Telegram air raid threat channels by sending each message to Ollama AI for it to evaluate, whether the message contains an actual threat to a city. Redirects the filtered messages to a Telegram chat.
- Deploy https://hub.docker.com/r/ollama/ollama
- Optionally: Deploy https://github.com/ollama-webui/ollama-webui/pkgs/container/ollama-webui
- Read through Modelfile. and use it to create a model via GUI or via CLI, or import it from Ollama Hub https://ollamahub.com/m/alert/ukraine-safety-assistant:latest (You can replace the city name and the base model, obviously.)
- Create .env file from .env.example
- Deploy Dockerfile (or image from
ghcr.io/alertua/air_raid_threat_reporter:latest) with volume/data, put yout.envthere. Alternatively, fill the Environment Variables. - The first launch must be interactive (
-it), to authenticate and create .session files in the volume folder. If the container is already created, then it'sdocker start -a ukraine-safety-assistant. If the container is already started, you can attach to it withdocker attach ukraine-safety-assistant