Repository navigation
[Skill idea] dart-run-quality-pipeline — orchestrate format + analyze + fix + test in one skill #12
Description
Environment
- Language model: Claude Opus 4.7
- Agent harness: Claude Code
- Project type: Production Flutter/Dart app with
Makefile-driven CI
Problem
The Dart roadmap has the right primitives — dart-run-static-analysis, dart-fix-static-analysis-errors (published), dart-collect-coverage, dart-run-build-runner — but no orchestrator that runs them in the canonical pre-commit / pre-push order with stop-on-fail semantics.
Every team I've seen ends up reinventing this pipeline (some as a Makefile target, some as a shell script, some as a CI step), which makes agent invocation inconsistent. A skill that encodes the canonical order makes the pipeline addressable as a single intent.
Proposed skill
Name: dart-run-quality-pipeline
Description (draft):
Run the canonical Dart quality pipeline (format → code generation → analyze → test) with stop-on-fail and report each step's status. Use before committing or pushing to verify a change won't break CI, with optional modes to skip code generation or scope tests to a path.
Modes:
- Full (default): format → build_runner → analyze → test
--ui: skip build_runner and tests (UI/widget changes only)--paths <path>: scope tests to a specific path
Behavior:
- Stop on first failing step
- Report exact error from the failing step
- Do not continue downstream steps on failure (avoids noisy compounding errors)
Why an orchestrator skill (not just chaining existing ones)
The mode logic and stop-on-fail semantics are the value. Without them, an agent running the existing skills in sequence will either (a) run all of them and bury the real error, or (b) ask the user mid-flow which is worse.
Real-world reference
We run this internally with three modes and it has measurably reduced "pushed broken CI" incidents. The --ui mode in particular saves ~90 seconds per UI iteration by skipping build_runner correctly.
Roadmap fit
Tools category, complementary to the existing analyze/coverage/build_runner skills.
Activity