claude-api

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly enables and instructs the agent to use WebSearch/WebFetch and MCP-driven browsing (see SKILL.md "When to Use WebFetch" and python/agent-sdk/README.md / patterns.md examples like "Open example.com and describe what you see" and references to shared/live-sources.md), which causes the agent to fetch and read arbitrary public web pages (untrusted third‑party content) that can influence its tool calls and next actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes runtime MCP server entries that fetch and execute remote packages (e.g., "npx @playwright/mcp@latest" and "npx @modelcontextprotocol/server-postgres"), which will download and run external code at runtime to provide agent tooling.