The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill uses administrative commands to modify system-level developer tool configurations. This behavior can lead to unauthorized system changes if the environment is not properly isolated.
Evidence: Usage of sudo xcode-select -s /Applications/Xcode.app/Contents/Developer and sudo xcodebuild -runFirstLaunch in references/upgrading-react-native.md for environment setup.
[EXTERNAL_DOWNLOADS]: The skill fetches resources from external repositories, including unified diff files and binary components.
Evidence: Downloads React Native template diffs from raw.githubusercontent.com/react-native-community/rn-diff-purge/ in SKILL.md and references/upgrade-helper-core.md.
Evidence: Downloads a binary Gradle wrapper (gradle-wrapper.jar) from the same community repository in references/upgrade-helper-core.md.
[COMMAND_EXECUTION]: The skill executes various shell commands to manage dependencies and build mobile applications, which could be exploited if inputs are manipulated.
Evidence: Extensive use of npm install, pod install, npx expo install, and native build tools like xcodebuild and react-native build-android across multiple files.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local project files and external diffs without sufficient sanitization or boundary markers, which could allow malicious instructions to influence the agent's behavior.
Ingestion points: Reads local configuration from package.json and external data from remote .diff files.
Boundary markers: The instructions do not define clear delimiters or provide 'ignore embedded instructions' warnings when processing these files.
Capability inventory: The skill has the ability to execute shell commands, perform network operations, and escalate privileges via sudo.
Sanitization: No evidence of validation or filtering for the content of processed project files or downloaded diffs.

upgrading-react-native