The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/_xcodebuildmcp_common.sh and scripts/bootstrap-xcodebuildmcp.sh scripts are designed to resolve the xcodebuildmcp tool. If the tool is not found locally, they utilize npx -y xcodebuildmcp@latest to download and execute the package from the NPM registry. This package is an open-source utility maintained by Sentry, a well-known service provider, and is used for its intended purpose of simulator automation. As the source is a well-known service, this is considered a standard dependency resolution.
[COMMAND_EXECUTION]: The skill includes several shell scripts (build-ios.sh, run-ios.sh, test-ios.sh, capture-screenshot.sh) that wrap xcodebuildmcp. These scripts execute shell commands to interact with xcodebuild and simctl for building, running, and testing iOS applications, as well as capturing screenshots of the simulator. This functionality is essential for the skill's primary purpose of design verification and audit.
[PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it is designed to process and audit untrusted user-provided SwiftUI source code and application screenshots.
Ingestion points: Untrusted data enters the agent context through the analysis of SwiftUI view code and simulator screenshots captured by capture-screenshot.sh.
Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded in the analyzed code or visual content.
Capability inventory: The skill possesses capabilities to execute shell commands for building (build-ios.sh), running (run-ios.sh), and testing (test-ios.sh) applications via xcodebuildmcp.
Sanitization: The provided shell scripts use standard bash safety flags (set -euo pipefail) but do not implement specific sanitization for external content before interpolation.

software-ios-design