The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands (curl and ssh) to perform best-effort connectivity tests.
[COMMAND_EXECUTION]: Variables derived from the user interview—such as host, ssh_port, ssh_user, and api_url—are interpolated directly into shell command strings (e.g., ssh ... -p "$SSH_PORT" "$SSH_USER@$HOST").
[COMMAND_EXECUTION]: There is no mention of input validation or sanitization for these user-provided strings, which creates a potential surface for command injection if a user provides inputs containing shell metacharacters (e.g., ;, |, or backticks).
[COMMAND_EXECUTION]: Ingestion points: User input is gathered via the 'Interview' procedure defined in SKILL.md and stored in $PLUGIN_DATA_DIR/config.json. Capability inventory: The skill utilizes mkdir, curl, and ssh shell capabilities. Sanitization: No sanitization logic is present in the prompt instructions to handle malicious input strings.

onboard