The Agent Skills Directory

[SAFE]: The skill logic is exclusively dedicated to content generation and educational workflows. No evidence of data exfiltration, credential harvesting, or malicious command execution was found.
[DATA_EXFILTRATION]: Security measures are included in the file storage rules to prevent the agent from interacting with sensitive system directories such as /etc, /usr, /bin, and /System. This ensures that generated learning materials are confined to user-approved locations.
[PROMPT_INJECTION]: The skill uses a feedback loop to personalize lessons. To protect against indirect prompt injection from untrusted data, the instructions define clear boundary markers and filtering rules, requiring the agent to disregard template text and focus only on verified user feedback sections.

dbs-learning