The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill uses Read, Glob, and Grep tools to scan the user's project files, detecting architecture patterns, deployment targets, and existing implementations (e.g., checking for existing StoreKit or analytics code). This data ingestion creates a surface for indirect prompt injection if the scanned files contain malicious instructions meant to influence the agent's code generation or tool usage.
[DYNAMIC_EXECUTION]: The app-icon-generator module programmatically generates a Swift script (generate-icon.swift) and then executes it using the swift interpreter to render icon assets. While this is a core part of its asset generation functionality, the runtime execution of generated source code is a sensitive pattern.
[COMMAND_EXECUTION]: The skill frequently uses the Bash tool to perform environmental audits, such as searching for print() statements to migrate to logging, verifying Info.plist keys, and checking project entitlements. It also uses the sips command-line utility to resize and process generated image assets.
[EXTERNAL_DOWNLOADS]: The generated templates for analytics and error monitoring include configuration instructions for integrating well-known third-party services, specifically TelemetryDeck, Firebase, and Sentry. These are recognized as established and trusted technology providers.

generators