Skip to content

Quick Fixes with Swift

Not every change needs the full spec-plan-execute pipeline. Arness Code offers three ceremony tiers that match process depth to change scope.

TierScopeFilesUse when
SwiftSmall1-8Bug fixes, config changes, small refactors, hotfixes, renames
StandardMedium4-15Feature additions, moderate refactors, API changes
ThoroughComplex15+New features, architectural changes, cross-cutting work

Arness detects complexity automatically when you run /arn-planning and suggests the right tier. You can always override.

/arn-code-swift

Or let /arn-planning route you there based on scope assessment.

Swift is a lightweight, pattern-aware workflow:

  1. Architectural assessment — Arness reads your codebase patterns and understands the conventions
  2. Inline plan — a brief description of what will change and why (no separate plan file)
  3. Execution — implements the change following your project’s patterns
  4. Verification — runs relevant tests
  5. Review — quick quality check

The output is a SWIFT_<name>.md change record and a SWIFT_REPORT.json — lightweight but sufficient for traceability. Every change gets documented, even the small ones.

/arn-code-standard

Standard bridges swift and thorough. It includes:

  • A spec-lite — lighter than a full feature spec but more structured than swift’s inline plan
  • Task-tracked execution — work is broken into tracked tasks
  • Review — quality check against patterns

The output is a STANDARD_<name>.md and STANDARD_REPORT.json.

For any change that involves UI components:

/arn-code-sketch

Sketch creates a preview of the UI change using your project’s actual framework and styling system. It works for web, CLI, and TUI applications. Use it during feature spec to validate a design, or during implementation to check your work.

Made a change outside the Arness pipeline? A 2am hotfix, an emergency patch, a quick manual edit?

/arn-code-catch-up

Catch-up examines recent commits and generates lightweight artifact records after the fact. This ensures the artifact trail remains complete even when the pipeline was bypassed.

When you run /arn-planning and describe your change, Arness assesses:

  • How many files are likely affected
  • Whether it crosses module/service boundaries
  • Whether it involves new patterns or follows existing ones
  • Whether there are test implications

Based on this, it suggests swift, standard, or thorough — and explains why. You always have the final say.