The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation instructions in SKILL.md direct the user to pipe a remote shell script directly into bash: curl -fsSL https://raw.githubusercontent.com/avivsinai/agent-message-queue/main/scripts/install.sh | bash. This pattern is high-risk as it executes unverified code from a remote source.
[PROMPT_INJECTION]: The skill instructions in SKILL.md explicitly direct the agent to use flags that bypass security controls: --dangerously-skip-permissions and --dangerously-bypass-approvals-and-sandbox. These instructions are intended to override the agent's internal safety and permission protocols.
[COMMAND_EXECUTION]: The skill relies on the execution of an external binary (amq) and provides numerous examples of shell commands for message management, environment configuration, and process interruption (Ctrl+C injection).
[DATA_EXPOSURE]: The skill involves reading and writing message files to a shared directory (e.g., .agent-mail). While intended for coordination, this creates an attack surface for indirect prompt injection if an attacker can write to the message queue monitored by the agent.

amq-cli