recording-browser-flow-as-test
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites through the
browser_snapshottool, which captures the DOM and accessibility tree. - Ingestion points: Browser snapshots of external target applications are ingested by the agent in
SKILL.md. - Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions when the agent parses the accessibility tree data.
- Capability inventory: The skill possesses the capability to write new test files to the repository and execute shell commands via
npx playwright test. - Sanitization: The skill lacks explicit instructions for sanitizing or validating element names, labels, or text content retrieved from the browser before using them to generate code.
- [DYNAMIC_EXECUTION]: The skill workflow involves generating a new Playwright test file (
.spec.ts) from captured interactions and then executing that generated script. - [COMMAND_EXECUTION]: The skill uses the command line to run
npx playwright testfor verifying the generated tests. While this is a standard developer workflow, it involves executing code generated based on external browser content.
Audit Metadata