It wouldn't be 2025 without mixing LLMs into everything -- I mean, tasteful integration of generative AI where appropriate. With that, anxious/excited to share my foray into prompting GitHub Models via GitHub Action: AI Inference Request. (There's "reasonable" rate-limits if you're keen to try/prototype with GitHub Copilot's free tier.) So far, we've been using it internally to generate in-line troubleshooting suggestions on job/step failure

if: ${{ failure() }}

, primarily to aid SDEs not too familiar with their test/build CI pipelines. Almost goes without saying, love to get your thoughts and feedback! I reckon more examples would go a long way.

Wow, that’s actually a really brilliant application of it

Sure thing, if there's spare time! Barely cobbled this together over the weekend (UTC+1 timezone 😅) so will get some polish by Wednesday's office-hours.