The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple Magento CLI commands (bin/magento) and shell commands (tail) to gather system information, status, and logs. While intended for diagnostics, these commands provide significant visibility into the underlying environment.
[DATA_EXFILTRATION]: The command bin/magento config:show is used to retrieve the entire system configuration. In Magento environments, this configuration often contains sensitive credentials, including payment gateway tokens, third-party API keys, and service passwords, which could be exposed to the agent context.
[PROMPT_INJECTION]: The skill reads external log files (var/log/system.log and var/log/exception.log) that are populated by application errors and exceptions. This creates a surface for indirect prompt injection, as an external attacker could potentially trigger specific errors that write malicious instructions into these logs to be processed by the agent.
Ingestion points: Reads from var/log/system.log and var/log/exception.log via the tail command (SKILL.md).
Boundary markers: None identified; log content is ingested without delimiters or safety instructions.
Capability inventory: Executes shell commands and Magento CLI tools (SKILL.md).
Sanitization: No evidence of sanitization or filtering of log content before processing.

magento-diagnostic