bx-ai-agents

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly configures external MCP servers (see "## MCP Servers (v3.0+)" with example urls and toolNames ["web_search","fetch_page"]), which allows the agent to fetch and interpret arbitrary web pages from third‑party sites and thus ingest untrusted public content that could influence actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The agent configuration includes an MCP server at https://api.example.com/mcp which is contacted at runtime to load remote tools (toolNames: ["*"]) that can inject behavior or execute code, so this URL is a runtime external dependency that can directly control the agent.