bgpt-paper-search

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

• Third-party content exposure detected (high risk: 0.85). The required workflow calls a remote MCP tool (search_papers) against a third-party server (bgpt.pro) that returns full-text-derived structured fields (methods/results/quality) which are outsider-authored paper content ingested into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill config directs runtime MCP calls to the remote SSE endpoint https://bgpt.pro/mcp/sse (and suggests running npx bgpt-mcp), which is invoked during agent runtime and returns data that is injected into the agent's context and thus can directly control prompts/outputs; the skill depends on that external server.