argent-ios-device-interact
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and act upon data retrieved from a physical device's user interface, creating a potential vector for malicious content embedded in apps or websites to influence the agent.
- Ingestion points: Data enters the agent's context through the output of the
describeandawait-ui-elementtools, which return the UI hierarchy and text content of the iPhone screen. - Boundary markers: There are no instructions provided to wrap UI content in delimiters or to apply "ignore embedded instructions" warnings when processing the screen state.
- Capability inventory: The skill uses a high-privilege set of interaction tools including
launch-app,open-url,gesture-tap,button, andkeyboard, allowing the agent to perform nearly any action on the device. - Sanitization: No sanitization, validation, or escaping logic is defined for the content scraped from the device before it is interpreted by the agent.
- [NO_CODE]: This skill consists entirely of markdown instructions and metadata without any accompanying scripts, executables, or automated configuration files.
Audit Metadata