verify-app
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the application under test, such as console logs and network responses, to generate proof reports. 1. Ingestion points: The skill collects evidence using the console, network-summary, and api commands. 2. Boundary markers: There are no explicit delimiters or instructions to ignore commands embedded in the evidence. 3. Capability inventory: The skill can execute browser actions like click, type, and upload via the verify.mjs script. 4. Sanitization: No sanitization or filtering of the application output is described before processing.
- [REMOTE_CODE_EXECUTION]: The skill executes local scripts that handle environment setup and tool invocation.
- [EXTERNAL_DOWNLOADS]: The setup command installs external dependencies and the Chromium browser from remote sources.
- [DYNAMIC_EXECUTION]: The skill mentions the use of eval to read application state within the browser context after user actions.
Audit Metadata