x-api

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime path calls X content via GET /api/search (and related endpoints) on https://twitter.fetcher.sh, where outsider-submitted search queries can cause the LLM to ingest arbitrary X post/account text returned by those queries.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly lists and configures an MCP server at https://twitter.fetcher.sh/mcp which is meant to be queried at runtime to expose MCP tools (search_endpoints, describe_endpoint, fetch_data) that directly change what actions/instructions an agent can execute, so this runtime URL controls agent behavior.