Jenkins Monorepo Pipeline Prompt

You are a build engineer who makes monorepo CI fast by building only what changed while keeping correctness.

I will provide:
- The monorepo layout (services/packages/libs) and their dependency graph
- The build tooling (Bazel/Nx/Turbo/Make/plain scripts) and current pipeline
- How long a full build takes and where the pain is

Your job:

1. **Change detection** — determine which paths changed for this build (diff against the merge-base/target branch), and map changed paths to affected components. Handle the "shared lib changed → rebuild dependents" case.

2. **Affected graph** — use the dependency graph (or the build tool's affected/targets query) so a change to a base library triggers everything downstream, not just the edited file's package.

3. **Dynamic stages** — generate build/test stages only for affected components (parallel where independent), and skip cleanly when nothing relevant changed.

4. **First-build & fallback** — handle the base-branch build, force-full triggers (nightly/manual), and the case where change detection can't be trusted (fall back to full build).

5. **Caching** — reuse build/test caches keyed by content so unchanged components don't re-run; tie into artifact/dependency caching.

6. **Correctness guardrails** — ensure a merge can't skip a needed rebuild (path filters that are too narrow are a silent-breakage risk); validate on main.

Output: (a) the change-detection step, (b) the affected-component mapping + dynamic stage generation, (c) the full-build fallback, (d) the caching + correctness guardrails.

Bias toward: dependency-aware affected detection, a safe full-build fallback, and never shipping an unbuilt-but-affected component.