bash-maintenance-orchestrator

name	bash-maintenance-orchestrator
description	Use when a Bash maintenance task needs coordinated testing, linting, and benchmarking instead of treating those loops as separate afterthoughts.

bash-maintenance-orchestrator

When to Use

Use this skill when the work spans more than one maintenance concern: correctness, coverage, lint hygiene, and performance.

Typical triggers:

• The user wants automated maintenance for a Bash framework or script suite.
• A change touches shared plumbing and needs regression tests plus lint plus performance checks.
• The task is broad enough that you need an execution order, not just one tool command.

Instructions

1. Classify the task before editing:
   • Correctness or regression risk.
   • Coverage gap.
   • Lint debt.
   • Performance hotspot.
2. Set the execution order:
   • Reproduce behavior first.
   • Add or tighten tests second.
   • Fix implementation third.
   • Run shellcheck fourth.
   • Run hyperfine last when performance matters.
3. Keep the loop narrow until the change proves out.
   • Target touched files and nearby tests first.
   • Expand to the full suite before finishing.
4. Require evidence for each claim:
   • A test result for behavior.
   • A lint rerun for warning cleanup.
   • A benchmark comparison for performance claims.
5. End with a maintenance summary:
   • What behavior is now protected?
   • Which warnings were removed or intentionally left?
   • What performance changed, if anything?

Decision rules:

• If correctness and performance conflict, protect correctness first and measure again.
• If lint cleanup creates risk without tests, pause and add coverage before proceeding.
• If a hotspot is not user-visible or operationally meaningful, avoid clever micro-optimizations.

Gotchas

• Do not run maintenance tools in a random order; that tends to bury the real regression source.
• Do not report “optimized” unless the benchmark loop actually improved.
• Do not close the task after a local narrow run if the repo has an established full-suite entrypoint.
• Tool output is not the same as judgment; prioritize the changes that materially improve the Bash maintenance loop.