recent-news-article-aggregator

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

• Third-party content exposure detected (high risk: 0.95). The required runtime workflow for search fetches and returns news article fields (e.g., title, description/snippet, url, source) from many public third-party publishers, so outsider-authored free text (article titles/summaries) is ingested into the agent’s LLM context via the tool’s JSON response.

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

• Potentially malicious external URL detected (high risk: 0.90). This skill explicitly instructs the agent at runtime to call the AgentPMT MCP endpoint (https://api.agentpmt.com/mcp/) to fetch live schema and "get_instructions" which are treated as authoritative and therefore can directly control prompts/instructions.