finalrun-test-runner

Pass

Audited by Gen Agent Trust Hub on Feb 27, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface because it processes test names and instructions retrieved from the FinalRun platform through tools like list_tests and list_test_suites. * Ingestion points: list_tests, list_test_suites, available_apps. * Boundary markers: Absent. * Capability inventory: run_test_locally, create_app_version, run_test_by_name_on_devices. * Sanitization: Absent.
  • [DATA_EXFILTRATION]: The create_app_version tool allows the agent to upload files from a local filePath to the vendor's infrastructure. While this is the intended mechanism for uploading application binaries for testing, it represents a capability that could be misused to target sensitive local files if the agent's logic is diverted.
  • [COMMAND_EXECUTION]: The skill manages the execution of mobile tests on local devices using system-level utilities such as adb and xcrun. These operations are performed through the run_test_locally and run_test_suite_locally MCP tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 27, 2026, 08:38 PM