dd-oci-integration
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads additional instructions and Terraform modules from official Datadog GitHub repositories. These operations target trusted organizations and repositories associated with the skill author.
- [COMMAND_EXECUTION]: The skill executes Terraform and the OCI CLI to manage infrastructure. It implements strong security controls, including mandatory user confirmation before resource modification and the use of restricted file permissions (
umask 077) for temporary plan files. - [CREDENTIALS_UNSAFE]: While the skill interacts with sensitive Datadog and OCI credentials, it employs sophisticated protection mechanisms. Secrets are passed to tools via stdin to prevent them from appearing in process lists (
ps) or shell history, and the instructions explicitly warn the user about protecting Terraform state files. - [INDIRECT_PROMPT_INJECTION]: The skill provides a mechanism to fetch and incorporate other skills from the author's official repository. It explicitly instructs the agent to treat these as reference material that cannot override existing safety protocols or user instructions.
- [REMOTE_CODE_EXECUTION]: The skill performs remote fetching and extraction of instructions from the author's GitHub repository. Because the source is a trusted organization, this pattern is used for legitimate cross-skill coordination rather than unauthorized code execution.
Audit Metadata