The Agent Skills Directory

[SAFE]: The skill is a defensive security tool designed to protect the user's environment. It implements pattern-based blocking and warning for dangerous operations.
[COMMAND_EXECUTION]: The skill configures a PreToolUse hook that executes a local shell script (guard/bin/check-dangerous.sh) whenever the Bash tool is invoked. This is the core mechanism for command validation.
[DATA_EXPOSURE]: The skill maintains an audit trail in a local directory ($NANOSTACK_STORE/audit.log) and reads session metadata (session.json) to enforce concurrency and environment-aware safety rules. This is legitimate functional behavior for an auditing tool.
[DYNAMIC_EXECUTION]: The Bash script uses the source command to load a local helper script (store-path.sh). This is standard modular shell scripting for the framework it belongs to.

guard