DEV Community

Brisbane Web Developer
Brisbane Web Developer

Posted on

Revolutionizing AI: How LLM like QwQ or Deepseek-R1 etc, Ollama, WireGuard, and Fabric Can Supercharge Your Workflow 

$HOME/.config/fabric/.env
======
DEFAULT_VENDOR=Ollama
DEFAULT_MODEL=qwq:latest
OLLAMA_API_URL=http://example.wireguard:12345
YOUTUBE_API_KEY=AIxb12345B0pQ8jL2nRt3xD5xH2wNabCd1dY-Z
Enter fullscreen mode Exit fullscreen mode 
$HOME/bin/ytext
======
#!/bin/bash

FABRIC_DIR="$HOME/.config/fabric"

url="$1"
options="${@:2}"

fabric -sp extract_wisdom -y "$url" $options && echo

session_name=$(echo "$options" | grep -oE -e "--session[ =][^ ]+" | sed -E 's/--session[ =]//')
if [ -n "$session_name" ];
then
  echo
  echo "Further prompts:"
  echo "fabric --session $session_name"
  echo
fi
Enter fullscreen mode Exit fullscreen mode 
# Store the transcript
ytext 'https://www.youtube.com/watch?v=example' --session example
Enter fullscreen mode Exit fullscreen mode 
# Investigate further
fabric --session example "What is ..."
Enter fullscreen mode Exit fullscreen mode

Top comments (1)

Collapse
 
p_o_26e854a54d851cd606f08 profile image
P O •

Nice combination. I'd keep Ollama bound to the tunnel address rather than 0.0.0.0, and verify the WireGuard AllowedIPs and firewall rules before exposing an API. A quick curl over the tunnel plus handshake counters catches most routing mistakes.