The Agent Skills Directory

[SAFE]: No malicious instructions, obfuscation, or security vulnerabilities were detected in the skill's logic or documentation.
[COMMAND_EXECUTION]: The skill utilizes shell access to perform xcodebuild build. This is an expected and necessary operation within an iOS development workflow to ensure that the generated SwiftUI prototypes compile correctly.
[DATA_EXPOSURE]: The skill reads from local project files such as .planning/APP.md to gather requirements. This access is limited to project-specific context and does not involve sensitive system-level credentials or private data.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a data ingestion surface through .planning/APP.md and interactive user questions. This information is used to parameterize the code generation. While there are no explicit boundary markers or sanitization steps mentioned for this external input, the risk is considered negligible as it operates within the developer's local environment and the output (SwiftUI code) is intended for direct developer review.
Ingestion points: .planning/APP.md and user responses via AskUserQuestion.
Boundary markers: Not explicitly defined for ingested text.
Capability inventory: File system writes (Write), code editing (Edit), and shell execution (Bash).
Sanitization: No explicit sanitization or escaping of external content is specified.

ui-prototyping