The Agent Skills Directory

[DYNAMIC_EXECUTION]: The scripts/generate_network_policy.py script contains a generate_script method that accepts arbitrary strings through the content argument and writes them directly into a shell script template. This allows the agent to generate executable bash files from untrusted or unverified text at runtime.
[PRIVILEGE_ESCALATION]: In scripts/generate_network_policy.py, the generate_script function calls file_path.chmod(0o755) on newly created files. Setting executable bits on dynamically generated content from user input is a security risk that facilitates the execution of potentially malicious code.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-defined networking requirements to generate manifests and scripts, creating a surface for injection attacks.
Ingestion points: User-provided descriptions and configurations passed to the generation script via SKILL.md triggers.
Boundary markers: Absent. There are no delimiters or instructions for the agent to ignore embedded shell commands within the networking descriptions.
Capability inventory: The skill has the ability to write files (Write tool), modify permissions (chmod in Python), and execute commands (Bash tool).
Sanitization: The generate_script function lacks any validation, escaping, or sanitization of the template content before writing it to a file.

managing-network-policies