The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard shell commands such as ls to list directory contents and grep to verify link integrity within the local vault. These operations are restricted to the $VAULT directory and are necessary for its file-management functionality.
[PROMPT_INJECTION]: The instructions do not contain any bypass attempts or overrides of agent safety guidelines. Instead, they implement strict internal rules (Red Flags) to prevent the agent from acting without explicit user consent.
[DATA_EXFILTRATION]: No network operations or external data transfer mechanisms were found. The skill operates entirely on the local file system within specified organizational folders.
[PRIVILEGE_ESCALATION]: All operations are performed within the user's workspace using standard permissions. There are no attempts to use sudo or modify system-level configurations.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests user input for naming folders and notes, it mitigates risk through a multi-stage confirmation process (Gate) in Stage 3, ensuring the user reviews all proposed file changes before they are executed.

area-creation