The Agent Skills Directory

[PROMPT_INJECTION]: The skill reads external log files which creates a surface for indirect prompt injection. Ingestion points: Log files located in the $MQL5_ROOT directory. Boundary markers: No explicit delimiters are used for the log content. Capability inventory: The skill uses Read, Bash, and Grep tools. Sanitization: The skill includes instructions to filter sensitive information before reporting results. Given the specialized nature of log reading, this is considered a standard operational risk.
[COMMAND_EXECUTION]: Local bash commands are utilized for path construction and reading file ends. These operations are static and do not involve untrusted input or privilege escalation.

log-reader