The Agent Skills Directory

[SAFE]: The skill provides configuration blueprints for system-standard logging tools including rsyslog, journald, and fluent-bit. The target domains (e.g., syslog.internal.example.com) and file paths (/var/log/auth.log) are appropriate for the intended use case.
[SAFE]: The Python AuditLogger implementation incorporates security-enhancing features such as hash-chaining (using hashlib.sha256) to ensure the integrity of log entries and detect potential tampering.
[SAFE]: The verify-log-integrity.sh script performs legitimate administrative functions by validating log file checksums and alerting via a webhook (ALERT_WEBHOOK_URL). It sets restrictive permissions (440) on its checksum files to protect integrity.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through the AuditLogger class and audit_trail decorator, which record application metadata and error strings. However, these inputs are serialized into structured JSON, and the skill does not perform any dangerous dynamic execution of the log content, making the risk profile acceptable for a logging utility.

audit-logging