The Agent Skills Directory

[SAFE]: The skill is designed to establish security best practices for the Codex environment, specifically promoting the use of sandboxes and request-based approval policies. The skill instructions explicitly warn against bypassing security approvals or using dangerous configurations.- [DATA_EXPOSURE_AND_EXFILTRATION]: The included shell script scripts/audit-codex-init.sh extracts configuration data for auditing purposes. However, it uses specific filtering logic to selectively print only non-sensitive configuration keys (e.g., approval_policy, sandbox_mode, and reasoning settings), effectively preventing the exposure of API keys or credentials stored in the configuration files.- [COMMAND_EXECUTION]: The skill utilizes standard system and application commands such as git status, codex --version, and codex doctor to verify the local environment. These are routine operations for a developer-focused agent skill and are executed within a local, read-only context during the audit phase.- [EXTERNAL_DOWNLOADS]: The skill references documentation from official OpenAI sources and well-known technical sites. These external references are informative and do not involve the download or execution of untrusted scripts or binary payloads.

laohan-chushicodex