FlakM (Nov 18 2024 at 14:00): FlakM (Nov 18 2024 at 14:00):Hi! I recently wrote about automating this process using markdown, jira-cli, and some help from LLMs using chatgpt-cli. It cut the friction, made planning more enjoyable, and saved me a lot of clicks. Thought someone might find it useful: https://flakm.com/posts/automate_boring_stuff/
Dustin (Nov 18 2024 at 17:43):Woah that's wayyy more trust than I have in llms :joy: But very cool
FlakM (Nov 18 2024 at 18:03):I hope someday my machine will be able to do similar things locally. I've played with Meta's models and the medium-size model, even quantized is decently capable. But from what I saw on shady Reddit posts it would take like three 3090s to get decent token/s.
FlakM (Nov 18 2024 at 18:11):But honestly I think there are probably a lot of practical use cases where data is not super secret and output is still quick to check by linters and humans. Glad you liked it!
Sukhdeep Brar (Nov 23 2024 at 23:02):Wow. That's way more automated than I assumed from the title. You have the LLM create a script that creates Jiras. I would have been excited from just having a Jira created from a problem statement. Thanks for sharing!
Andrei Jiroh Halili (Dec 05 2024 at 05:21):Not a Jira user myself, but I'll be happy to cook up ChatGPT CLI and GitHub/GitLab CLI.
FlakM (Dec 05 2024 at 14:45):@Andrei Jiroh Halili Let us know how it works
Last updated: Dec 12 2024 at 15:17 UTC