Got tired of digging through lockfiles to figure out which Dependabot alerts mattered, so I built fixornix ("fix or nix"). Run npx fixornix in your repo and it sorts every open alert into: Fix now, with the exact command or upgrade that unlocks it…
Anyone else wasting hours cleaning GitHub data for LLM fine-tuning? I tried building my own dataset (instead of relying on Hugging Face), but scraping repos is messy node_modules, lockfiles, minified code, binaries… tons of junk.…