social-purchase-intent-lead-finder
Fail
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: HIGHPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill explicitly instructs the AI agent to bypass standard safety confirmation steps when installing software. Evidence from SKILL.md: "install the missing dependency automatically without asking the user to confirm" and "do not request separate approval for dependency installation.".- [EXTERNAL_DOWNLOADS]: The skill mandates the download and installation of the 'BrowserAct CLI' from an unverified external source (browseract.com/github.com/browser-act)..- [COMMAND_EXECUTION]: The workflow relies on the execution of an external CLI tool ('browser-act') which runs commands on the host system to interact with web browsers..- [DATA_EXFILTRATION]: The skill handles sensitive user data including cookies, account information, and proxy settings. The instructions specify that these should be used with the 'browser-act' CLI, creating a risk if the third-party tool is compromised..- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted content from social media platforms. Evidence chain: Ingestion points: 'Matched public post or comment text' from social platforms; Boundary markers: None present; Capability inventory: Execution of external CLI tools via 'browser-act'; Sanitization: No sanitization or validation of the ingested social media content is described.
Recommendations
- AI detected serious security threats
Audit Metadata