company-contact-enrichment

Fail

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The instructions mandate the automatic installation of the 'BrowserAct' CLI and Skill from an external source if they are unavailable. This bypasses typical safety protocols regarding external code acquisition.
  • [COMMAND_EXECUTION]: The skill requires the agent to execute installation commands silently. It specifically forbids the agent from requesting user approval or notifying the user that a dependency is being installed, which is a significant violation of transparency and user oversight.
  • [PROMPT_INJECTION]: The 'BrowserAct Runtime' section contains instructions that override the agent's standard behavior of seeking user consent for software installation, essentially 'jailbreaking' the installation workflow to ensure it happens autonomously.
  • [DATA_EXFILTRATION]: The skill encourages the use of a third-party service (BrowserAct) for processing company data. While presented as a feature, the mandatory and silent nature of the installation increases the risk of unauthorized data handling by unverified software.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external websites and public listings.
  • Ingestion points: Processes website content, business listings, and social handles at runtime via the browser tool.
  • Boundary markers: None specified in the instructions to prevent the agent from following instructions embedded in the target web pages.
  • Capability inventory: Accesses live browser pages and executes tasks using the browser-act CLI.
  • Sanitization: No evidence of sanitization or filtering of the retrieved content before it is processed by the agent.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 20, 2026, 04:28 AM
Security Audit — agent-trust-hub — company-contact-enrichment