The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The shell script scripts/generate-rbac.sh uses a heredoc to execute Python code. It interpolates the shell variables $ORG_CHART_FILE and $OUTPUT_FILE directly into Python string literals. A maliciously crafted filename containing single quotes and Python commands (e.g., '); import os; os.system('...) could lead to arbitrary code execution when the script is run.
[REMOTE_CODE_EXECUTION]: The scripts/generate-rbac.sh script relies on the PyYAML library without ensuring it is installed, creating a dependency on the host environment's configuration.
[PROMPT_INJECTION]: The skill includes an ingestion surface for untrusted data in scripts/generate-rbac.sh, which reads and processes an organizational chart (org-chart.yaml).
Ingestion points: The scripts/generate-rbac.sh script reads an external YAML file defined by the user.
Boundary markers: The script uses yaml.safe_load() to parse the input file.
Capability inventory: The script generates policy files that define the authorization rules for users and roles within the application.
Sanitization: Employs yaml.safe_load() to mitigate risks associated with unsafe YAML deserialization.

authorization-security