kogiqa-mcp-browser-automation

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The required workflow is browser automation that can ingest page content at runtime via the kogiqa_get_content tool (HTML/text extracted from arbitrary URLs/pages, i.e., outsider-authored web content), which the agent may then place into its LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill starts the MCP server with "npx kogiqa-mcp@latest", which at runtime fetches and executes remote code from the npm package (see https://www.npmjs.com/package/kogiqa-mcp), so the skill relies on an external package that is downloaded and run during execution.