The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the official oci and oci-cli packages using pip. These are well-known libraries provided by Oracle for interacting with their cloud infrastructure and do not represent a security risk when used for their intended purpose.
[COMMAND_EXECUTION]: The skill executes standard system commands such as openssl to generate RSA keys and the oci command-line interface to verify account connectivity. These operations are necessary for the skill's primary function of setting up API signing keys.
[DATA_EXFILTRATION]: The skill interacts with sensitive configuration files located in ~/.oci/. It correctly implements security best practices by applying restrictive file permissions (chmod 600) to both the private key and the configuration file. It uses clear placeholders for user-specific identifiers (OCIDs), preventing the hardcoding of actual credentials.
[DYNAMIC_EXECUTION]: The skill generates a diagnostic Python script (validate_oci_config.py) to verify the local configuration. The generated code is static, transparent, and performs legitimate checks such as verifying file existence and permissions before attempting an authentication call via the official SDK.
[INDIRECT_PROMPT_INJECTION]: The skill has a minimal attack surface for indirect prompt injection as it primarily processes local configuration data and direct user input for OCIDs. It does not ingest untrusted data from external network sources during its setup workflow.
Ingestion points: Processes user-provided OCIDs and the content of the local ~/.oci/config file in SKILL.md.
Boundary markers: Not explicitly used, but the input is constrained to specific configuration fields.
Capability inventory: Includes file system write/read and shell command execution via the Bash tool.
Sanitization: Utilizes the official OCI SDK's built-in configuration validation methods.

oraclecloud-install-auth