qa-explore
Warn
Audited by Gen Agent Trust Hub on Jul 1, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructs the agent to search for and read sensitive local files, specifically mentioning
.env.localandbrain/local/test-credentials.md, to retrieve authentication tokens and credentials for use in the browser session. - [COMMAND_EXECUTION]: The skill executes arbitrary shell commands defined in the project's
package.json(such aspnpm dev,npm run dev, oryarn dev) and utilizes theagent-browsertool to manipulate the UI. - [PROMPT_INJECTION]: The skill ingests data from untrusted or external sources including browser accessibility trees, console logs, and project-specific documentation files (e.g.,
CLAUDE.md). This creates an indirect prompt injection surface (Category 8) because the agent processes this content to make decisions without the use of boundary markers or sanitization.
Audit Metadata