The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various command-line utilities to perform its primary function. It uses terraform (or tofu) for cloud infrastructure provisioning, gcloud for discovering GCP organizations, folders, and projects, and curl for secure interaction with the Datadog API. These commands are consistent with the skill's stated purpose of managing cloud integrations.
[EXTERNAL_DOWNLOADS]: The skill downloads content from external sources. It retrieves additional skill modules from the official datadog-labs GitHub repository (https://codeload.github.com/datadog-labs/agent-skills/tar.gz/refs/heads/main) and fetches delegate principal emails from the Datadog API. These downloads originate from trusted domains associated with the skill's author.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources that could potentially contain malicious instructions.
Ingestion points: Untrusted data enters the context via .env files, gcloud CLI output (e.g., project and folder names), and HTTP responses from the Datadog API.
Boundary markers: The instructions include a warning to the agent that downloaded skills are reference material only and must not override safety rules or user instructions.
Capability inventory: The skill possesses capabilities to execute shell commands (terraform, gcloud, curl), read environment files, and write Terraform configuration and plan files to the local disk.
Sanitization: There is no explicit sanitization or validation logic applied to strings retrieved from the GCP environment or API responses before they are interpolated into the generated Terraform HCL or displayed to the user.

dd-gcp-integration