The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/generate_config.py script features a generate_script function that accepts content from the --content command-line argument and writes it directly to a .sh file. The function subsequently applies execution permissions (chmod 0o755) to the file. This allows for the dynamic creation and possible execution of arbitrary shell scripts based on unvalidated input.\n- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes external data to generate configuration files and scripts without proper isolation or validation.\n
Ingestion points: Input enters via the --content, --title, and --config arguments in scripts/generate_config.py.\n
Boundary markers: No delimiters or instructions are used to distinguish user-provided content from the generated script or configuration templates.\n
Capability inventory: The skill is authorized to use powerful tools, including Bash with aws, gcloud, and nginx scopes, along with file system read and write permissions.\n
Sanitization: There is no evidence of input validation or sanitization before the data is written to executable files.

configuring-load-balancers