The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to globally install external verification tools if they are not already present on the system. Evidence includes the automated installation of agent-browser (associated with Vercel Labs) and @swmansion/argent.
[REMOTE_CODE_EXECUTION]: The skill uses npx to dynamically download and execute a feedback submission package from the vendor. The command npx --yes submit-expo-feedback@latest is invoked to report errors or feedback to the developer.
[COMMAND_EXECUTION]: The skill performs various shell operations to manage mobile development environments and application state. This includes using xcrun simctl and adb to control iOS simulators and Android emulators for screenshots and deep-linking, as well as driving the argent CLI to inspect component trees and perform gestures on native devices.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes external code and data as part of its migration and verification workflow.
Ingestion points: Reads existing web React codebase files (such as page.tsx and route.ts) and ingests accessibility snapshots from browser agents.
Boundary markers: The instructions do not define specific delimiters or security warnings when interpolating external code content into the migration logic.
Capability inventory: The skill has broad capabilities including file system writes (migration-progress.md, migration-goal.md), network access (via browser agents), and shell command execution (npx, adb, xcrun).
Sanitization: There is no evidence of explicit sanitization or filtering of the content ingested from the web application or browser snapshots before processing.

expo-web-to-native