The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill explicitly instructs the agent to bypass standard security boundaries by running commands "unsandboxed" and requesting "escalated permissions" to access host resources.
[COMMAND_EXECUTION]: The skill executes a JavaScript companion script (agy-companion.mjs) using the node runtime via a shell command string.
[CREDENTIALS_UNSAFE]: The instructions specify that the tool requires access to a sensitive "OAuth token file" on the host system to function correctly.
[INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface for indirect prompt injection or command injection due to the following:
Ingestion points: User-provided questions are accepted as input for the --prompt flag in SKILL.md.
Boundary markers: Absent. The instructions require passing the user's question "verbatim" without delimiters.
Capability inventory: The skill possesses the capability to execute shell commands via node as defined in SKILL.md.
Sanitization: Absent. There are no requirements or logic provided to escape or sanitize the user input before it is interpolated into the shell command.

agy-ask