The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a custom runtime protocol to execute its internal utility scripts (e.g., structure_state.py, freshness_state.py). These executions are performed using subprocess.run with shell=False, and the runtime includes checks to ensure executed scripts reside within the skill's own directory.
[SAFE]: Filesystem operations are governed by security checks that prevent path traversal and ensure that symlinks do not resolve to locations outside the target repository root.
[SAFE]: No evidence of network communication, credential harvesting, or prompt injection attempts was found. The skill maintains state locally in dedicated JSON files within the repository.

map-codebase