---

name: "swiftui-code-review" description: "Review a SwiftUI change for actor isolation, observation, and main-actor UI updates. Use for a SwiftUI code review. Do not use for a non-Swift repository or to silence isolation with unchecked Sendable."

SwiftUI Code Review Skill for Copilot

Copilot skill swiftui-code-review for SwiftUI. The description decides when Copilot loads it. Keep this copy under .github/skills.

When to run

Run this review for a SwiftUI view, a model, or a concurrency change.

When to leave it

Stop for a non-Swift repository. Do not change the deployment target or the Swift tools version unless the user asked.

Steps

1. Read the Swift tools version and the deployment target. An API used below that target is a high finding. @unchecked Sendable added in the diff is a high finding unless a comment proves the invariant.
2. Check UI updates. A published property written from a non-main-actor task is a high finding.
3. Search for try? and ! on decoded or optional values. Each one that hides a failure is a finding.
4. Check task lifetime. .task is cancelled with the view. A plain Task { } with no stored handle and no cancel() is a finding. A task that never checks cancellation during a long loop is a finding.
5. Note tests. Run the project's test action only if the user asked. Otherwise write not run.

Output

Findings with severity, file, line, the isolation or availability issue, and a fix.

## Findings
- high | path:line | what is wrong | what to do instead

## No findings
Say "No findings" and list the files you opened. A search you skipped is not a clean result.

## Example
Input: `ProfileView.body` starts `Task { await model.load() }`, so every redraw starts another load.
Output: high | ProfileView.swift:11 | the load is an unstructured task in `body` | use `.task { await model.load() }`. A `Task` inside a button action is not this finding.

Do not flag the valid pattern named in the standards. Flag the mistake next to it.

Stop

Stop when the user asks you to disable strict concurrency to get a green build.

Severity

high is a bug a caller can hit or a secret that leaks. medium is a broken contract that needs a specific input. low is a cleanup that does not change behaviour.