combine-code-review

What This Skill Does

The combine-code-review skill acts as a specialized static analysis assistant for Apple's Combine framework. It is designed to identify common architectural pitfalls, performance bottlenecks, and dangerous memory patterns that frequently arise in reactive Swift codebases. By leveraging a comprehensive checklist—including verification of [weak self] in closures, correct usage of AnyCancellable storage, and appropriate error handling via tryMap or catch—this agent provides actionable feedback to developers. It specifically targets high-risk areas like operator misuse, improper usage of assign(to:on:), and subject management, ensuring your reactive streams are both memory-safe and predictable.

Installation

To integrate this skill into your environment, run the following command within your terminal while the OpenClaw agent is active: clawhub install openclaw/skills/skills/anderskev/combine-code-review

Use Cases

This skill is essential when refactoring legacy code that uses RxSwift or early Combine implementations. It is best utilized during pull request reviews, feature architectural design phases, and debugging sessions involving unexpected UI behavior or memory growth. Developers should invoke this skill when the codebase contains imports of Combine, heavy use of @Published properties, or chains involving complex transformations like flatMap and switchToLatest.

Example Prompts

1. "Review this view model for potential retain cycles and ensure all subscriptions are handled correctly: [paste code here]"
2. "I am getting a memory leak in my Combine pipeline. Can you audit my sink closures and AnyCancellable set usage?"
3. "My publisher chain keeps terminating early on network errors. How can I use catch or flatMap to keep the main stream alive?"

Tips & Limitations

To get the most out of the skill, always provide context regarding the ownership model (e.g., is the publisher inside a View, ViewModel, or Service?). The agent relies on specific keyword detection; ensure you provide enough surrounding code to allow it to trace AnyCancellable lifecycles effectively. Note that this skill is a static analysis tool; it cannot execute your code or detect runtime race conditions that require local environment debugging. Always cross-reference the suggested fixes with the provided documentation files like memory.md or operators.md to ensure a deep understanding of the reactive principles being applied.