CI/CD Pipeline Designer

CI/CD Pipeline Designer audits your existing delivery pipeline, identifies the stages consuming the most time and producing the least signal, and returns a prioritized roadmap to fix them. It applies cost-to-information ordering — cheap checks like linting and type analysis run before slow integration suites — so every failure surfaces at the earliest, cheapest point. It also designs flaky-test quarantine protocols, dependency caching strategies, deployment gates, and rollback triggers, all adapted to whichever CI platform and language stack you specify.

A buyer opens the skill and describes their situation: a Node.js monorepo on GitHub Actions where the full pipeline takes 22 minutes, three tests fail non-deterministically on every other run, and the team deploys manually to production because nobody trusts the pipeline. The skill asks four targeted questions about stack, goal, constraints, and audience, then produces a structured audit and action plan calibrated to that specific setup.

The output covers distinct modules: a bottleneck table ranking stages by duration and failure rate, a reordered stage sequence with rationale for each move, a caching configuration block for npm dependencies keyed to package-lock.json hash, a quarantine job definition that isolates the three flaky tests without blocking the main build, and a canary deployment template with automatic rollback triggers based on error-rate thresholds. Each section is copy-paste ready or formatted as a checklist with owner and verification fields, depending on the output format chosen.