Refactor to ultra-specialized agents with clear handoffs - #2
Conversation
## Summary - Rename commands to drop `swift-` prefix: /feature, /plan, /test, /build - Add new commands: /review, /modernize - Split TCA into architect (design) + engineer (implementation) - Add new agents: swift-code-reviewer, swift-modernizer - Rename agents for clarity: swift-architect, swift-engineer, swiftui-specialist, etc. - Optimize model usage: Opus for planning, Sonnet for implementation, Haiku for builds - Add 2025 context to all agents - Remove programming-swift from defaults (load on-demand) - Planning agents are now READ-ONLY - Each agent has clear identity statement and handoff table 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Code Review: Ultra-Specialized Agent RefactorSummaryThis is an excellent refactor that transforms the Swift Engineering plugin into a well-architected, specialized agent system. The separation of concerns, clear handoff patterns, and model optimization are particularly well done. The implementation demonstrates strong software architecture principles and should significantly improve the user experience. Overall Assessment: ✅ Approved with minor suggestions 🎯 Strengths1. Exceptional Architecture Design
2. Excellent Documentation
3. Strong Quality Patterns
4. Well-Designed Commands
🔍 Issues & SuggestionsCRITICAL Issues: 0Important Issues: 21. [Consistency] Missing Bash tool in swiftui-specialist agentFile: Issue: The Why it matters: The agent may need to run git commands, check file status, or perform other shell operations. All other implementation agents include Bash. Fix: -tools: Read, Write, Edit, Glob, Grep
+tools: Read, Write, Edit, Glob, Grep, BashJustification: Consistency with other implementation agents (swift-engineer, tca-engineer, swift-test-creator). 2. [Documentation] Self-evaluation question inconsistencyFiles: Multiple agent files Issue: Most agents use "Have I done the best possible work I can?" but some use more specific questions like "Is the build clean?" or "Have I followed the TCA Design exactly?" Why it matters: The generic question is less actionable than specific criteria. More specific self-evaluation leads to better quality control. Examples:
Suggestion: Update all agents to use specific, measurable self-evaluation criteria rather than the generic question. Suggestions: 41. [Enhancement] Add test coverage guidanceContext: Neither Suggestion: Consider adding a coverage expectation section to the plan file format: ## Test Strategy
- **Target coverage:** 80%+ for reducers, 60%+ overall
- **Key behaviors to test:** [List]
- **Edge cases:** [List]This helps set expectations and measure test completeness. 2. [Documentation] Clarify "programming-swift" on-demand loadingFiles: Multiple agent files mention loading Issue: The guidance is consistent but could be more explicit about HOW to load it. Suggestion: Add a brief note in agent files: ## programming-swift Usage
Load `programming-swift` skill ONLY when:
- Verifying obscure Swift syntax
- Checking language semantics (e.g., actor isolation rules)
- Resolving compiler errors related to language features
**To load:** Add "Read the programming-swift skill for [specific question]" in your prompt.3. [Workflow] Consider adding error budget to agentsContext: Suggestion: Consider adding an error budget section to implementation agents: ## Error Handling
- Attempt fixes up to 2 times
- On third failure, escalate to appropriate agent or user
- Document all failures in handoff notesThis prevents infinite loops and clarifies when to ask for help. 4. [Performance] Plan file update frequencyContext: Agents are instructed to update the plan file "before returning to main." Consideration: For long-running implementations, the plan file might not reflect in-progress status if the agent hasn't completed. Suggestion: Consider adding guidance for progress updates during long-running work: ## Progress Updates
For implementations taking >5 file changes:
- Update handoff notes in the plan file after each major milestone
- This helps other agents if handoff happens mid-implementation🎨 Code Quality AssessmentBest Practices: ✅
Maintainability: ✅
Testing Considerations: ✅
🔒 Security Review✅ Good Security Practices
No Security Issues Found ✅📊 Performance Considerations✅ Strong Performance Design
Potential OptimizationContext size management: Multiple agents reading the same plan file could lead to context bloat if files get large. Consider adding guidance: ## Plan File Best Practices
- Keep handoff notes concise (3-5 sentences per agent)
- Archive old handoff entries after completion
- Use bullet points over paragraphs🧪 Test Coverage AssessmentTest Plan in PR Description: ✅ Comprehensive The test plan covers:
Suggestion: Consider adding integration tests:
📝 Documentation QualityREADME.md: ⭐⭐⭐⭐⭐Strengths:
Minor enhancement: Could add a "Quick Start" section for new users: ## Quick Start
1. Install Sosumi MCP server (required)
2. Run `/feature` to create your first Swift feature
3. Choose "fully automated" for hands-off workflow
4. Review generated code in `Features/<FeatureName>/`Agent Documentation: ⭐⭐⭐⭐⭐Exceptional quality:
🎯 RecommendationsHigh Priority
Medium Priority
Low Priority (Nice to Have)
Final VerdictThis PR is ready to merge. The two important issues are minor consistency fixes that don't block functionality. The suggestions are enhancements for future consideration. Why this is excellent work:
Impact: This refactor transforms the plugin from a basic workflow tool into a sophisticated agent orchestration system. It should significantly improve success rates for Swift development tasks. Commit RecommendationOnce the two consistency issues are addressed (or accepted as-is), this is ready to merge. Great work on this refactor! 🎉 Review conducted by Claude Sonnet 4.5 |
| 3. Rebuild | ||
| 4. If fixed → continue | ||
| 5. If still failing after 3 attempts → write to handoff notes, ask user | ||
| 5. If still failing after 3 attempts → write to handoff notes, hand off to specialist |
There was a problem hiding this comment.
Should the file specifically state which specialists?
|
|
||
| ## On Completion | ||
|
|
||
| 1. **Update the plan file** with the TCA Design section |
There was a problem hiding this comment.
Won't this agent need access to Write tools to do this?
|
|
||
| Before returning to main: | ||
|
|
||
| 1. **Create the plan file** at `docs/plans/<feature>.md` |
There was a problem hiding this comment.
Won't this agent need write tools to do this?
|
|
||
| **BOUNDARY:** If issues found, hand back to implementation agent. Otherwise, hand off to @swift-test-creator. | ||
|
|
||
| *If semi-guided:* "Review complete. Ready for testing?" |
There was a problem hiding this comment.
It should ask if the user wants tests to be written. If the answer is no, proceed to asking the user if they want to document (phase 8)
## Summary - Created `modern-swift` skill with Swift 6.2 concurrency essentials - Created `modern-swiftui` skill with iOS 17+ SwiftUI patterns - Added `modern-swift` to all 11 agents for consistent concurrency patterns - Added `modern-swiftui` to 4 UI-focused agents (swift-ui-design, swiftui-specialist, swift-code-reviewer, swift-modernizer) - Updated README.md with new skills documentation ## Skills Content **modern-swift:** - async/await (no completion handlers) - @mainactor (no DispatchQueue.main) - Actor isolation (no locks) - Sendable conformance - TaskGroup (no DispatchGroup) **modern-swiftui:** - @observable (replaces ObservableObject) - @State (replaces @StateObject) - @bindable (replaces @ObservedObject) - environment(_:) (replaces environmentObject) - onChange(of:initial:_:) new signature - task(priority:_:) for async work ## Why Without essential Swift knowledge loaded by default, agents might use deprecated patterns. These condensed skills ensure all agents follow modern Swift 6.2 and iOS 17+ best practices. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Code Review: Swift Engineering Plugin RefactorReviewer: @swift-code-reviewer (Claude Sonnet 4.5) SummaryThis is an excellent refactoring that transforms the Swift Engineering plugin into a well-architected, ultra-specialized agent system. The separation of concerns, clear handoff boundaries, and model optimization (Opus for planning, Sonnet for implementation, Haiku for builds) are particularly well done. The addition of Overall Assessment: ✅ Ready to merge with minor suggestions for improvement. Strengths[PRAISE] Agent ArchitectureThe split between planning (read-only, Opus) and implementation (write-enabled, Sonnet) agents is brilliant:
[PRAISE] Skills DesignThe plugins/swift-engineering/skills/modern-swift/SKILL.md
plugins/swift-engineering/skills/modern-swiftui/SKILL.md
[PRAISE] Command DesignThe plugins/swift-engineering/commands/feature.md:46-223
[PRAISE] Documentation QualityThe README.md provides excellent overview: plugins/swift-engineering/README.md:1-177
Important Issues[IMPORTANT] iOS 26.0 Future DatingFiles: Multiple agents and README Examples:
Fix: Update to realistic versions: # Before
**Platform:** iOS 26.0+, Swift 6.2+, Strict concurrency
# After
**Platform:** iOS 17.0+, Swift 6.0+, Strict concurrencyImpact: This affects all agent context sections. iOS 17 is appropriate for [IMPORTANT] Inconsistent Plan File Status ItemsFile: Multiple commands and agents Example 1 - feature.md: plugins/swift-engineering/commands/feature.md:23-32
- [ ] UI design (@swift-ui-design)
- [ ] Architecture (@swift-architect)
- [ ] TCA design (@tca-architect) — if TCA
- [ ] Implementation (@tca-engineer or @swift-engineer)
- [ ] Views (@swiftui-specialist)
- [ ] Code review (@swift-code-reviewer) — optional
- [ ] Tests (@swift-test-creator)
- [ ] Build verified (@swift-builder)
- [ ] Documentation (@swift-documenter) — optionalExample 2 - swift-architect.md: plugins/swift-engineering/agents/swift-architect.md:100-109
- [ ] UI design analysis (@swift-ui-design) — if applicable
- [x] Planning complete (@swift-architect)
- [ ] TCA design (@tca-architect) — if TCA
- [ ] Core implementation (@tca-engineer or @swift-engineer)
- [ ] SwiftUI views (@swiftui-specialist)
- [ ] Code review (@swift-code-reviewer) — optional
- [ ] Tests (@swift-test-creator)
- [ ] Build verified (@swift-builder)
- [ ] Documentation (@swift-documenter) — if neededFix: Standardize the status checklist across all files. Consider creating a single source of truth in a shared template. [IMPORTANT] Missing Error Handling GuidanceFiles: Agent implementation guides Current state in swift-engineer.md: plugins/swift-engineering/agents/swift-engineer.md:64-68
### Error Handling
- Domain-specific error types
- Typed throws (Swift 6.2)
- Handle errors at appropriate boundariesSuggestion: Add concrete examples: // Domain-specific error types
enum UserProfileError: Error {
case invalidEmail
case networkFailure(Error)
case unauthorized
}
// Typed throws (Swift 6.0+)
func loadProfile() async throws(UserProfileError) -> UserProfile {
// Implementation
}
// Handle at boundaries
func refreshProfile() async {
do {
let profile = try await loadProfile()
self.profile = profile
} catch let error as UserProfileError {
self.error = error.localizedDescription
}
}Suggestions[SUGGESTION] Agent Identity ConsistencyFiles: All agent files Best example: plugins/swift-engineering/agents/swift-architect.md:13-16
You are **@swift-architect**, an expert iOS/Swift software architect.
**Mission:** Design Swift feature architectures that are maintainable, testable, and follow Apple best practices.
**Goal:** Produce a comprehensive plan that implementation agents can follow without ambiguity.Weaker example: plugins/swift-engineering/agents/swift-builder.md:11-16
You are **@swift-builder**, an expert in Xcode builds and compiler error resolution.
**Mission:** Build the project and resolve compiler errors efficiently.
**Goal:** Achieve a clean build with no errors or warnings.Suggestion: Ensure all agents have equally strong identity/mission/goal statements. The swift-builder example is actually good, but ensure consistency across all 11 agents. [SUGGESTION] Plan File Location ConventionFiles: Commands and agents Current: plugins/swift-engineering/commands/feature.md:18
All agents share state via a plan file at `docs/plans/<feature-name>.md`:Suggestion: Add explicit naming conventions: ## Plan File Naming
Format: `docs/plans/<feature-name>.md`
Conventions:
- Use kebab-case: `user-profile.md` (NOT `UserProfile.md` or `user_profile.md`)
- Match feature name: `UserProfile` feature → `user-profile.md`
- Descriptive: `shopping-cart.md`, not `cart.md` or `feature-1.md`[SUGGESTION] Skill Loading OptimizationFiles: All agents Example - swift-architect.md: plugins/swift-engineering/agents/swift-architect.md:6
skills: modern-swift, ios-hig, composable-architecture, sqlite-dataSuggestion: The architect loads plugins/swift-engineering/agents/swift-architect.md:88-92
## programming-swift Usage
Load `programming-swift` skill ONLY when:
- Verifying obscure Swift syntax
- Checking language semantics (e.g., actor isolation rules)
- This skill is 37K+ lines - use sparinglyCould apply similar guidance for other large skills. [SUGGESTION] Test Coverage ExpectationsFiles: Suggestion: Add test coverage guidelines to ## Test Coverage Expectations
Aim for meaningful coverage, not 100% for its own sake:
**Always test:**
- Public API surface
- Business logic and domain rules
- Error handling paths
- State transitions (especially in TCA reducers)
**Consider skipping:**
- Simple property getters/setters
- UI layout code without logic
- Generated code
**Coverage targets:**
- Critical features: 80%+ coverage
- Standard features: 60%+ coverage
- UI components: Focus on behavior, not rendering[SUGGESTION] Modernize Command Missing DetailsFile: Current: Lists migrations but no prioritization ## Migration Priority
When multiple modernization opportunities exist:
1. **Safety first:** Concurrency issues (completion handlers → async/await)
2. **Platform alignment:** SwiftUI adoption (ObservableObject → @Observable)
3. **Maintainability:** TCA adoption (if appropriate)
4. **Cleanup:** Deprecated APIs, old patterns
**Why this order:**
- Concurrency bugs are the hardest to diagnose
- Modern SwiftUI reduces boilerplate significantly
- TCA is architectural (bigger decision)
- Cleanup has lowest riskSecurity Considerations✅ No Security Issues Found
Performance Considerations✅ Well Optimized
One enhancement suggestion: ## Agent Token Usage (Estimated)
| Agent | Typical Usage | Why |
|-------|---------------|-----|
| @swift-architect | 15-25K tokens | Large context (skills + planning) |
| @tca-engineer | 8-15K tokens | Medium context (implementation) |
| @swift-builder | 3-8K tokens | Small context (Haiku model) |Test Coverage AssessmentStatus: Observation: This is a plugin/configuration change, not executable code, so traditional unit tests don't apply. However, consider: [SUGGESTION] Add Integration Test Plan: ## Manual Testing Checklist
Before release, verify:
- [ ] `/feature` command executes full workflow
- [ ] Planning agents (Opus) cannot write files (verify with test command)
- [ ] TCA architect → TCA engineer handoff preserves context
- [ ] `@swift-builder` uses Haiku model (check logs)
- [ ] Skills load correctly (modern-swift, modern-swiftui)
- [ ] Plan file format is consistent across agents
- [ ] MCP server (Sosumi) lookup works in all agents
Test with sample feature: "Add a user profile screen with TCA"Code Quality & Best Practices✅ Excellent Structure
✅ Naming Conventions
[SUGGESTION] Add Version/ChangelogObservation: No version tracking for the plugin itself. Suggestion: Add to README.md: ## Version
**Current:** 2.0.0
### Changelog
#### 2.0.0 (2025-12-20)
- Major refactor: Ultra-specialized agents with clear handoffs
- Split TCA into architect + engineer
- Added modern-swift and modern-swiftui skills
- Renamed commands (dropped `swift-` prefix)
- Added @swift-code-reviewer and @swift-modernizer
- Model optimization (Opus/Sonnet/Haiku)
#### 1.0.0 (Initial)
- Original swift-development pluginVerdict✅ APPROVED - Ready to MergeBlocking Issues: None Recommended Pre-Merge Actions:
Optional Enhancements (can be follow-up PRs):
Summary of Review Findings
Excellent work on this refactor! 🎉 The agent specialization and clear handoff patterns are exactly what a complex workflow plugin needs. This will significantly improve the developer experience for Swift/iOS development with Claude Code. Review conducted by Claude Sonnet 4.5 using swift-code-reviewer patterns |
## Summary - Created `swift-common-patterns` skill with architecture patterns from swift-agents-plugin - Created `swiftui-common-patterns` skill with SwiftUI patterns from swift-agents-plugin - Added pattern skills to agent frontmatter (9 agents updated) - Updated README.md skills table - Agents stay lean - all code examples in skills, not in agents ## Skills Content **swift-common-patterns** (architecture patterns): - DefaultProvider protocol pattern - ConcreteResolvable pattern (type-safe token resolution) - Actor-isolated managers - Model-ViewModel-View pattern - Environment-based dependency injection - Modern networking architecture - Error handling patterns **swiftui-common-patterns** (UI patterns): - MVVM with @observable (iOS 17+) - NavigationStack patterns - Performance optimization (lazy loading, equatable views, debounced search) - UIKit interoperability (UIViewRepresentable, UIViewControllerRepresentable) - Accessibility patterns (VoiceOver, Dynamic Type, accessibility actions) - Async operation patterns (task modifier, refreshable, background tasks) - View composition patterns (ViewBuilder, custom modifiers) ## Agent Updates **Added swift-common-patterns to:** - swift-architect - tca-architect - swift-engineer - tca-engineer - swift-test-creator - swift-modernizer **Added swiftui-common-patterns to:** - swiftui-specialist - swift-ui-design - swift-code-reviewer - swift-modernizer ## Benefits - Agents stay lean (under 200 lines each) - Rich code examples available via skills - Patterns extracted from production swift-agents-plugin - Agents automatically load skills via frontmatter - Single source of truth for patterns 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Code Review: Ultra-Specialized Agents RefactorSummaryThis is an excellent refactor that dramatically improves the plugin's architecture through clear separation of concerns, optimized model usage, and well-defined agent boundaries. Verdict: ✅ Ready to merge with minor suggestions for enhancement. Strengths🎯 Architecture & Design[PRAISE] Agent Specialization Model
[PRAISE] Handoff Model
[PRAISE] Plan File as Shared State [PRAISE] Documentation Quality Code Quality Analysis✅ Best Practices Followed
RecommendationsImportant (Address Before or Shortly After Merge)1. [IMPORTANT] iOS/Swift Versions 2. [IMPORTANT] Handoff Chain Validation
3. [IMPORTANT] Dependency Audit 4. [QUESTION] MCP Server Availability Suggestions (Nice to Have)
Specific File Highlights✅ Excellent Files
Security, Performance, Testing✅ Good Practices
Test CoveragePR includes manual test plan. Consider adding automated validation of agent configurations and handoff chains. Final ThoughtsThis refactor is a significant improvement. The agent boundary design, model optimization, and handoff model are exceptional and could serve as patterns for other agent-based systems. The TCA architect/engineer split shows smart recognition that architectural reasoning benefits from Opus, while implementation is well-served by Sonnet. Recommendation: Merge with confidence. The "Important" items are minor and could be addressed in follow-up PRs. Great work! 🎉 Reviewed by: Claude Sonnet 4.5 |
|
@claude iOS 26 does exist. Just update all of the files to specify iOS 26+.
I agree with this. Make the changes and ensure that swiftui-specialist shows both paths.
Sure
Fallback to programming-swift skill is fine, lets put that in.
Do it
Do it
Don't worry about this one.
Yes tell them that is a target of theirs, but if its impossible not to get too hung up on it.
Don't worry about that
Show me what these might look like Also, I left some code comments on the PR. Fix these up if you think they are relevant please. Open a PR with your changes. |
|
Claude finished @johnrogers's task in 4m 58s —— View job Review Feedback Implemented ✓All requested changes have been implemented and pushed to the Completed Tasks
Changes SummaryAgent Updatesswift-architect.md:
tca-architect.md:
swift-ui-design.md:
swift-builder.md:
swiftui-specialist.md:
feature.md (workflow):
New Documentationdocs/smoke-tests.md:
Smoke Tests OverviewThe smoke test document provides:
These tests validate that the plugin configuration matches the design specification and that agent behavior follows intended patterns. Note: All changes were pushed directly to the |
- Add dependency audit criteria to swift-architect - Add fallback to programming-swift when Sosumi unavailable - Add error triage guidance to swift-builder - Expand test strategy template in swift-architect - Add context budget guidance for Opus agents - Update swiftui-specialist to show both handoff paths (with/without review) - Add Bash tool to swiftui-specialist - Add Write/Edit tools to swift-architect and tca-architect for plan file updates - Update feature workflow to ask about tests after review/views - Add comprehensive smoke test documentation Co-authored-by: John Rogers <johnrogers@users.noreply.github.com>
Code Review: Swift Engineering Plugin RefactorSummaryThis is a well-architected major refactor that successfully transforms the plugin into a highly specialized, modular system with clear agent boundaries and handoff patterns. The design demonstrates thoughtful separation of concerns, appropriate model optimization (Opus for planning, Sonnet for implementation, Haiku for builds), and comprehensive documentation. Overall Assessment: ✅ Approve with minor suggestions Strengths1. Excellent Agent Boundary Design ⭐The separation between planning (read-only, Opus) and implementation (write-capable, Sonnet) agents is brilliant:
2. Comprehensive Skill System 📚The new skills are exceptionally well-designed:
3. TCA Architecture Split 🎯Separating
4. Plan File as Shared State 📋The
Issues & Suggestions[IMPORTANT] 1. Inconsistent Agent Tool PermissionsFile: Issue: The agent metadata declares Problem: This creates a contradiction between the tool permissions and the behavioral instructions. Fix: Remove Write and Edit from tools list for read-only agents (swift-architect, tca-architect, swift-ui-design) Impact: Medium — Could lead to accidental file modifications during planning phase. [IMPORTANT] 2. Missing Error Escalation GuidanceFile: Suggestion: Add explicit escalation criteria for when to hand off vs. when to escalate to user after multiple failed attempts. Security Review ✅Findings: No security concerns detected
Performance Considerations ⚡Findings: Well-optimized
Code Quality & Best Practices ⭐Excellent Patterns
Final Verdict✅ APPROVED with minor fixes recommended This refactor significantly improves the plugin's architecture and usability. The agent specialization, clear handoffs, and comprehensive skills make this a production-ready system. Recommendation: Merge after fixing tool permissions for read-only agents. Reviewer: Claude Sonnet 4.5 |
1 participant
Summary
Major refactor of the Swift Engineering plugin to create ultra-specialized agents with clear boundaries and handoff patterns.
Key Changes
swift-prefix):/feature,/plan,/test,/build/review,/modernizetca-architect(design, Opus) +tca-engineer(implementation, Sonnet)swift-code-reviewer,swift-modernizerprogramming-swiftremoved from defaults - loaded on-demand onlyAgent Overview
@swift-ui-design@swift-architect@tca-architect@tca-engineer@swift-engineer@swiftui-specialist@swift-test-creator@swift-builder@swift-documenter@swift-code-reviewer@swift-modernizerWorkflow
Test plan
/featurecommand triggers correct agent sequence/plancommand creates plan without implementation/testcommand creates and runs tests/buildcommand runs build verification/reviewcommand executes code review/modernizecommand handles legacy migrationsFull Implementation Plan
Swift Engineering Plugin Refactor Plan
Goals
swift-prefix:/feature,/plan,/test,/build@swift-architect,@swift-engineer, etc.@agent-namereferences with WHY rationale in commandsprogramming-swiftfrom defaults, specify when to use itCommand Changes
Rename Commands (drop
swift-prefix)/swift-feature/featurecommands/feature.md/swift-plan/plancommands/plan.md/swift-test/testcommands/test.md/swift-build/buildcommands/build.md/reviewcommands/review.md/modernizecommands/modernize.mdCommand Content Updates
Each command must:
@agent-namesyntaxAgent Changes
Rename/Refactor Agents
swift-plannerswift-architectswift-ui-designswift-ui-designswift-tcatca-architecttca-engineerswift-coreswift-engineerswift-swiftuiswiftui-specialistswift-testingswift-test-creatorswift-buildswift-builderswift-docsswift-documenterswift-code-reviewerswift-modernizerAgent Template Structure
For PLANNING agents (opus model, read-only):
For IMPLEMENTATION agents (sonnet/haiku model, can write):
Agent Specifications
Planning Agents (Opus, READ-ONLY)
@swift-architect@swift-ui-design@tca-architectImplementation Agents (Sonnet/Haiku)
@tca-engineer@swift-engineer@swiftui-specialist@swift-test-creator@swift-builder(Haiku)@swift-documenter@swift-code-reviewer@swift-modernizerFiles Created
Commands:
commands/feature.mdcommands/plan.mdcommands/test.mdcommands/build.mdcommands/review.md(NEW)commands/modernize.md(NEW)Agents:
agents/swift-architect.mdagents/tca-architect.mdagents/tca-engineer.mdagents/swift-engineer.mdagents/swiftui-specialist.mdagents/swift-test-creator.mdagents/swift-builder.mdagents/swift-documenter.mdagents/swift-code-reviewer.md(NEW)agents/swift-modernizer.md(NEW)Files Deleted
Commands:
commands/swift-feature.mdcommands/swift-plan.mdcommands/swift-test.mdcommands/swift-build.mdAgents:
agents/swift-planner.mdagents/swift-tca.mdagents/swift-core.mdagents/swift-swiftui.mdagents/swift-testing.mdagents/swift-build.mdagents/swift-docs.mdFiles Updated
agents/swift-ui-design.md(2025 context, descriptions OR mockups, READ-ONLY)README.md(new workflow documentation)Skill Assignment (Optimized)
@swift-architect@swift-ui-design@tca-architect@tca-engineer@swift-engineer@swiftui-specialist@swift-test-creator@swift-builder@swift-documenter@swift-code-reviewer@swift-modernizerNote:
programming-swiftremoved from all defaults. Agents know to load it when:🤖 Generated with Claude Code