oci-compute
Fail
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes the Fn CLI installation script from the official Fn Project GitHub repository using a piped shell command sequence (curl | sh).
- [COMMAND_EXECUTION]: Provides example
cloud-initconfigurations that include directives for creating users with passwordless sudo privileges and modifying system firewall rules. - [COMMAND_EXECUTION]: Uses the OCI CLI to modify sensitive local configuration files, specifically the Kubernetes config (~/.kube/config), to store cluster access credentials.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading external configuration files (e.g., cloud-init.yaml) and interpolating their contents into shell commands used for infrastructure deployment.
- Ingestion points: Reads user-defined
cloud-init.yamlfiles via theread_filetool. - Boundary markers: No specific delimiters or "ignore instructions" headers are used to wrap the content of these files before they are processed by the agent.
- Capability inventory: The skill has access to the
shelltool and can execute commands to create cloud instances, deploy functions, and modify network settings. - Sanitization: There is no evidence of validation or sanitization of the content of the configuration files before they are used in command execution.
- [EXTERNAL_DOWNLOADS]: Downloads binaries and installation scripts from the
raw.githubusercontent.comdomain to support the OCI Functions development environment.
Recommendations
- HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/fnproject/cli/master/install - DO NOT USE without thorough review
Audit Metadata