Skills
Navigate
skills/revitalb-port/skills-registry-demo/verifying-in-browser/Gen Agent Trust Hub

verifying-in-browser

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes npm run dev to initiate a local development server. This pattern is standard for development-oriented tasks but relies on the integrity of the local package.json scripts.
  • [INDIRECT_PROMPT_INJECTION]: The skill exposes the agent to indirect prompt injection by processing external data from the browser environment.
  • Ingestion points: The agent reads data from browser_console_messages, browser_network_requests, and browser_snapshot (SKILL.md).
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat browser data as untrusted or to ignore instructions embedded within that data.
  • Capability inventory: The agent has capabilities to execute shell commands (npm run dev) and navigate/interact with web environments.
  • Sanitization: No validation or sanitization mechanisms are described for the content ingested from the browser tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 01:53 PM