skills/laststance/skills/qa-ios/Gen Agent Trust Hub

qa-ios

Pass

Audited by Gen Agent Trust Hub on Apr 29, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard xcrun simctl commands. These commands are used to manage simulator state, launch applications, and capture log streams, which is appropriate for a QA automation tool. It also involves reading system crash reports from ~/Library/Logs/DiagnosticReports/ and writing session logs to /tmp/qa-ios-session/ to report app stability.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it ingests and processes accessibility (AX) trees from external iOS applications via the mcp__ios-simulator__ui_describe_all tool. 1. Ingestion points: Accessibility tree data retrieved in SKILL.md (Phases 1, 2, 3, 4, and 6). 2. Boundary markers: None present; the agent is instructed to directly parse labels and roles from the AX JSON. 3. Capability inventory: The agent has access to Bash for shell execution and various UI interaction tools like ui_tap and ui_type across all scripts. 4. Sanitization: There is no evidence of sanitization or filtering of the UI element labels before the agent uses them to determine its next actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 29, 2026, 03:13 AM