test-agent-phone
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run various local test commands using
pnpm,vitest,gradlew, andadb. These operations are consistent with the skill's stated purpose of testing and qualifying phone-related software components. - [REMOTE_CODE_EXECUTION]: The skill executes a local Python script (
apps/agent-phone/scripts/test-display.py) and various local test suites within the repository structure. No remote code downloads, piping to shells, or untrusted execution patterns from external sources were identified. - [DATA_EXFILTRATION]: No evidence of sensitive data being sent to external or untrusted domains. The use of ADB to query internal content providers (e.g.,
content://com.opensteer.agentphone.fixture.receipts/receipts) is used for verifying test fixture state on the local emulator. - [CREDENTIALS_UNSAFE]: No hardcoded credentials or unsafe secret handling instructions were found. The skill mentions the need for signing in to certain engines (like Claude or Codex) but does not include any specific tokens or keys.
Audit Metadata