AI-Native Team Research

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly directs the agent to search and extract live content from public, user-generated sources (e.g., X/Twitter via use-agently, Hacker News/site:news.ycombinator.com, Brave Search, Tavily, and Firecrawl) and to read/synthesize those results into a briefing, so untrusted third‑party content could materially influence the agent's actions and enable indirect prompt injection.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly requires and instructs the user to "Fund your wallet with USDC on Base" and to use the CLI flag --pay to authorize payments for agent calls; it also references use-agently balance and that agent calls require funds. Those instructions indicate an integrated crypto wallet/payment flow (on Base, with USDC) and an ability to authorize payments—i.e., signing/sending transactions. This is a specific crypto/ blockchain payment capability (not just generic HTTP or browser automation), so it meets the Direct Financial Execution criteria.