The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes shell script snippets for environment detection (checking for CI/GitHub Actions variables) and for polling iOS/macOS simulator readiness. These commands are local diagnostics used to ensure the execution environment is prepared for testing.- [DATA_EXPOSURE]: The skill reads local project metadata from files such as Package.swift, *.xcodeproj, and .gstack/track. This access is scoped to identifying the project type and build platforms required for routing the testing request to the correct internal executor.- [INDIRECT_PROMPT_INJECTION]: The skill evaluates untrusted data from the local file system (project configuration) and user-provided verbs to make routing decisions. The ingestion points include build settings and directory structures, while the capability inventory is limited to invoking internal sub-skills. There are no explicit boundary markers for this data, but the sanitization is inherent in the deterministic routing logic which only maps specific inputs to a predefined routing table.

e2e-route