Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly a company-research helper, but it automatically allows browser fallback for auth-gated sites like LinkedIn without requiring clear user consent.
Install only if you are comfortable with a company-research skill that may move from Exa search into browser-based access when results are insufficient or content is gated. Before using it, prefer public-source research and require explicit approval before any logged-in LinkedIn or other authenticated browser access.
The description is broad enough to trigger the skill for many generic company-, market-, or competitor-related requests, increasing the chance of unintended activation. While not directly dangerous on its own, overbroad invocation can route unrelated tasks into workflows that perform external searches or browser fallback behavior the user did not specifically intend.
The skill explicitly instructs a browser fallback for auth-gated LinkedIn or other content, which expands behavior beyond the declared Exa-only workflow and can lead the agent to access authenticated sessions or protected content. In a research skill, this is risky because it may cause unintended use of a user's logged-in browser context, with privacy, policy, and account-impact implications.
The browser fallback for auth-gated content lacks any requirement to warn the user that the agent may interact with logged-in sites such as LinkedIn or other protected pages. This is dangerous because it can expose private account context, trigger account actions or tracking, and bypass the user's expectation that the skill is limited to Exa search.
No suspicious patterns detected.
