← 返回 Skills 市场
Review Ios
作者
Kevin Anderson
· GitHub ↗
· v1.0.1
· MIT-0
118
总下载
0
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install review-ios
功能描述
Comprehensive iOS/SwiftUI code review with optional parallel agents
安全使用建议
This skill looks like a legitimate iOS code-review guide, but there are two things you should confirm before installing or running it:
1) Why is disable-model-invocation set to true? The SKILL.md expects the agent to invoke the model, spawn subagents (Task), and load other skills (Skill), which typically requires model/tool invocation. Confirm with the author whether the skill is intended to run without model access or whether this flag is an error — as-is it may be non-functional or hide a dependency.
2) The skill instructs loading many beagle-ios:* skills at runtime but does not declare them in metadata. Ask which of those skills will actually be loaded and verify they come from trusted owners; loading external skills expands the runtime attack surface.
Also note: the skill will run git/grep and may run swiftlint on your repository, so it needs read access to the codebase and (optionally) the swiftlint binary. If you run it, ensure the beagle-ios dependencies exist and are trusted, and that you understand the consequences of allowing the agent to read repository files.
能力评估
Purpose & Capability
Name/description (iOS/SwiftUI code review) align with the actions in SKILL.md: scanning repo files, running SwiftLint, and producing review Issues. The requested operations (git, grep, swiftlint, reading .swift files) are appropriate for a code-review skill.
Instruction Scope
Instructions require running repository commands (git diff, grep), re-reading files by path:line, running SwiftLint, and producing structured Issues — these are within expected scope. However, SKILL.md requires spawning subagents via the Task tool and loading many other skills (beagle-ios:*). Those actions presume the skill can invoke the model and the Skill/Task tooling to run subagents; yet the skill metadata sets disable-model-invocation: true, which prevents normal model-driven behavior. That's a functional inconsistency that will prevent the described runtime flow or hide that the skill depends on other skills' runtime behavior.
Install Mechanism
Instruction-only; no install/spec or external downloads. Lowest-risk install mechanism.
Credentials
No environment variables, credentials, or config paths requested. The skill operates only on repository files and local binaries (git, grep, swiftlint), which is proportionate for code review.
Persistence & Privilege
always is false and the skill explicitly disables model invocation (reduce privilege). The SKILL.md asks to load other skills at runtime but those dependencies are not declared in the registry metadata — this is a transparency issue (the skill will depend on other skills that could broaden privileges when loaded).
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install review-ios - 安装完成后，直接呼叫该 Skill 的名称或使用
/review-ios触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
**Adds strict "hard gate" requirements before code review can proceed.**
- Introduces four mandatory gates: explicit file scope, linter enforcement, review protocol loading, and evidence for critical/major findings.
- The review process must not start until all gates are satisfied and skills load in the correct order.
- If no Swift files are in scope or if gates are unmet, review is stopped without reporting issues.
- Requires referencing the verification protocol for all findings, noted in the output summary.
- Strengthens evidence requirements for every critical/major issue (must observe and match the real file context).
v1.0.0
- Initial release of the review-ios skill for comprehensive iOS/SwiftUI code review.
- Detects changed Swift files, verifies linter status with SwiftLint, and identifies used iOS technologies.
- Loads relevant technology-specific code review skills automatically or in parallel via `--parallel`.
- Enforces a strict single-pass review policy (all issues across all categories at once, scoped to the diff).
- Provides a detailed review process including verification, output formatting, and post-fix checks.
- Disallows reporting issues covered by project linters and adheres to Apple framework idioms.
元数据
常见问题
Review Ios 是什么？
Comprehensive iOS/SwiftUI code review with optional parallel agents. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 118 次。
如何安装 Review Ios？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install review-ios」即可一键安装，无需额外配置。
Review Ios 是免费的吗？
是的，Review Ios 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Review Ios 支持哪些平台？
Review Ios 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Review Ios？
由 Kevin Anderson（@anderskev）开发并维护，当前版本 v1.0.1。
推荐 Skills