sweetlink

[Skill Scanner] Installation of third-party script detected SweetLink's functionality aligns with its stated purpose and will be valuable for local automation and agent-driven browser tasks. However, it intentionally grants high privileges: access to authenticated browser sessions, DOM/console/network telemetry, and the ability to simulate user input. Installing a custom TLS CA and running a long-lived daemon increases local trust and supply-chain risk. The package as described contains no explicit obfuscated or clearly malicious code in the provided text, but its design amplifies the impact of a compromised binary or a malicious caller. Recommended controls before use: obtain binaries from signed/verified releases, audit the package source code, restrict local access to the daemon (firewall, OS-level permissions), avoid running on machines with sensitive credentials unless necessary, and require explicit, documented authorization controls for callers. LLM verification: The SweetLink skill's documented capabilities are consistent with its stated purpose (connecting an AI agent to a real browser via a local daemon). There is no direct evidence in the provided documentation of obfuscated or overtly malicious code or third-party exfiltration. However, the tool inherently exposes highly sensitive data (cookies, localStorage, DOM, console/network logs) to any agent or process that can communicate with the local daemon. Key risks: (1) trusting a local CA (trust-ca) m