The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system commands including terraform and tofu for infrastructure management, the oci CLI for tenancy verification, and standard utilities like awk, grep, and sed for configuration parsing.
[EXTERNAL_DOWNLOADS]: Fetches the official Datadog OCI Terraform module from github.com/DataDog/oracle-cloud-integration. This repository belongs to a trusted organization and is the official source for this integration.
[DYNAMIC_EXECUTION]: Uses the shell eval command within environment variable loading logic to check for the presence of Datadog keys in .env files. This is used for local configuration management.
[DATA_EXFILTRATION]: The skill reads sensitive configuration files, specifically ~/.oci/config and .env files, to retrieve OCIDs and authentication keys. This data is used locally to configure Terraform and validated against official Datadog API endpoints (api.datadoghq.com). The skill includes explicit logic to identify and suppress the printing of sensitive fields like pass_phrase to the agent's output.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests and processes data from external sources:
Ingestion points: API responses from Datadog (via curl) and output from terraform state list.
Boundary markers: None explicitly defined in the shell snippets, though the agent is instructed to show plan outputs to the user for manual review.
Capability inventory: The skill has significant capabilities including file system writes (.tf files, tfplan), network requests via curl, and resource provisioning via terraform apply.
Sanitization: The skill uses precise parsing (e.g., awk and grep with specific patterns) to extract only expected fields, which limits the risk of processing malicious instructions embedded in tool outputs.

dd-oci-integration