The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and executes CLI tools including eas-cli, agent-device, and argent from the NPM registry. These tools originate from the vendor (Expo) or established ecosystem contributors (Callstack, Software Mansion) and are used to manage the remote simulation environment.
[COMMAND_EXECUTION]: Leverages a wide range of shell commands to manage native builds, interact with the remote simulator daemon, and establish network tunnels for live development.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from the simulator, such as UI snapshots and system logs, which represents a potential attack surface for indirect prompt injection.
Ingestion points: Data is ingested through agent-device snapshot -i, and the sim_api helper which retrieves device logs (/logs) and crash reports (/crashes).
Boundary markers: The instructions do not specify the use of delimiters or specific "ignore" markers when interpolating simulator data into the agent's context.
Capability inventory: The skill maintains high-privilege capabilities including arbitrary shell command execution via the Bash tool, local file system writes for artifacts and logs, and network operations via expo start --tunnel and curl.
Sanitization: No automated sanitization or validation of the simulator-provided content is described before processing by the agent.

eas-simulator