If you've ever built a feed, search ranker, or notification system, you've reinvented this pipeline shape: fetch candidates, hydrate metadata, filter the ineligible, score what's left, pick top K, log impressions async. This skill scaffolds that six-stage pattern (popularized by xAI's open-sourced X algorithm) into working code for your stack. The real value is in the trade-offs it surfaces: single score versus multi-action prediction, online versus batch, filter ordering. It won't write your ML model, but it will wrap it in proper plumbing so you're not debugging why cache writes are blocking user responses at 3am.
npx -y skills add wshobson/agents --skill recsys-pipeline-architect --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
kubesphere/kubesphere
supercent-io/skills-template