The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to execute Node.js scripts via the command line to perform HTTP operations. The scripts/request.js file allows saving output to a file using the --output flag, which utilizes path.resolve without sanitizing or restricting the destination path. This could allow an agent to write to or overwrite arbitrary files on the system if it provides a path like .bashrc or other sensitive configurations.- [INDIRECT_PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection because it is designed to ingest and return raw data from untrusted external URLs.
Ingestion points: Both scripts/request.js and scripts/fetch-json.js use the fetch API to retrieve data from user-supplied URLs.
Boundary markers: Absent. The scripts print the raw response body or JSON content directly to standard output without any delimiters or instructions to the agent to treat the content as untrusted data.
Capability inventory: The skill can perform diverse network operations (GET, POST, PUT, DELETE, PATCH) and has file system write access via the --output flag in the request script.
Sanitization: Absent. There is no validation, filtering, or escaping of the content fetched from remote servers before it is presented to the agent's context.- [DATA_EXFILTRATION]: The skill's core functionality is to send data to external URLs. This capability can be leveraged to exfiltrate sensitive information, such as environment variables or the contents of local files, by passing them as the body of a POST or PUT request to a server controlled by an attacker.

http-client