claude-api

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most URLs are official Anthropic/Claude docs and SDK repos, but the MCP server URL https://api.githubcopilot.com/mcp/ looks like a typosquatting/unknown domain (possible impersonation of GitHub/Copilot) and could be used to host or distribute malicious tool artifacts.

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

• Potentially malicious external URL detected (high risk: 0.80). The Managed Agents examples include runtime-mounted Git repositories and MCP server URLs (e.g. "https://github.com/owner/repo", "https://github.com/org/repo", "https://api.githubcopilot.com/mcp/", "https://mcp.slack.com/mcp") that are supplied in session/agent creation requests and therefore are fetched/attached at runtime and can deliver code or content the agent will read or execute, so they are runtime external dependencies capable of controlling prompts or executing code.