superset-mcp

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly says the agent "sends [the API key] as a Bearer token", which instructs embedding a user's secret API key verbatim in Authorization headers (a direct secret-exfiltration pattern).

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

• Potentially malicious external URL detected (high risk: 1.00). The skill requires runtime calls to https://api.superset.sh/api/v2/agent/mcp (and to fetch the MCP server card at https://api.superset.sh/.well-known/mcp/server-card.json) which expose tools that launch coding agents and run commands/terminals on hosts, so these URLs directly control prompts and can cause remote code execution.