paralel-search

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

• Third-party content exposure detected (high risk: 1.00). The skill (SKILL.md and app.py) exposes the agent to untrusted third-party content by calling the Parallelsearch MCP endpoint (https://search-mcp.parallel.ai/mcp) via the web_search_preview tool (web searches returning search results) and the web_fetch tool (fetching content from arbitrary HTTP/HTTPS URLs), which the agent reads and parses as part of its workflow and therefore could allow indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill uses a runtime MCP endpoint (https://search-mcp.parallel.ai/mcp) via Client.call_tool to fetch search results and arbitrary web content (through web_fetch) which is returned and intended for LLM consumption, so external content from that URL can directly influence agent prompts at runtime.