The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing administrative commands through several system utilities, including the AWS CLI (aws), Redis CLI (redis-cli), MySQL (mysql), and RabbitMQ administrator (rabbitmqadmin). It also suggests direct modification of the NGINX configuration file located at /etc/nginx/nginx.conf.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to the Cloudflare API using curl to manage zone settings. This interaction targets a well-known technology service and is consistent with the skill's stated purpose.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection due to the interpolation of data into executable shell commands.
Ingestion points: User-provided parameters such as instance-id, zone-id, group name, and REDIS_API_KEY are interpolated into shell command templates in SKILL.md.
Boundary markers: No delimiters or specific instructions to ignore embedded commands are present in the usage patterns.
Capability inventory: The skill possesses high-privilege capabilities including subprocess execution (AWS, Redis, MySQL, RabbitMQ) and network access via curl.
Sanitization: There is no evidence of input validation or sanitization for parameters before they are passed to the shell.

arch-scalability