The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The skill instructs the user to execute a remote script via curl https://sdk.cloud.google.com | bash. This is a critical security risk as it executes unverified code from a remote server directly in a shell environment.\n- [PROMPT_INJECTION] (HIGH): The skill exhibits a significant Indirect Prompt Injection surface (Category 8). It performs high-privilege cloud operations (deployment, infrastructure management) using data ingested from untrusted sources like Dockerfiles and project configurations.\n
Ingestion points: Project files and configuration files processed during deployment and optimization tasks.\n
Boundary markers: None; there are no delimiters or instructions to treat external data as untrusted.\n
Capability inventory: Broad cloud management capabilities including wrangler deploy, gcloud run deploy, and docker build.\n
Sanitization: Absent; the skill lacks any mechanisms to sanitize or validate external content before use in high-privilege commands.\n- [COMMAND_EXECUTION] (MEDIUM): The skill uses potentially dangerous shell operations, such as creating files via cat and running build processes that could be exploited if environment variables or input files are compromised.\n- [EXTERNAL_DOWNLOADS] (MEDIUM): The skill installs the wrangler CLI and Google Cloud SDK from sources not explicitly included in the provided Trusted External Sources list.

devops