The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (HIGH): The skill uses npx to download and run the @aiconnect/confidant package from the npm registry. This package and its organization are not listed in the Trusted External Sources, posing a supply chain risk.
[REMOTE_CODE_EXECUTION] (HIGH): Runtime execution of remote npm packages via npx allows for arbitrary code execution on the agent's system. An attacker could potentially compromise the package to gain control over the agent.
[CREDENTIALS_UNSAFE] (HIGH): The skill's primary purpose is to handle high-value secrets such as API keys, passwords, and tokens. Directing these sensitive items through an unverified third-party tool creates a significant risk of credential exposure or theft.
[COMMAND_EXECUTION] (MEDIUM): The skill relies on shell command execution (npx ...). If parameters like the URL or label are sourced from untrusted user data, it could lead to command injection.
[DATA_EXFILTRATION] (MEDIUM): While the tool claims to be secure, its ability to send data to arbitrary URLs provided in the fill command creates a vector for data exfiltration if the agent is tricked into sending secrets to an attacker-controlled endpoint.

confidant