openai-docs
Warn
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to add MCP servers (e.g.,
codex mcp add openaiDeveloperDocs). - [PRIVILEGE_ESCALATION]: In the 'If MCP server is missing' section, the skill explicitly instructs the agent to 'retry with escalated permissions' if a command fails due to sandboxing or permissions issues.
- [CREDENTIALS_UNSAFE]: The skill provides instructions for the agent to modify the user's
~/.claude/settings.jsonfile. This is a sensitive configuration file that often contains API keys and other credentials for MCP servers. - [METADATA_POISONING]: The 'OpenAI product snapshots' section includes misleading information regarding 'gpt-oss' models and 'Codex' capabilities that are inconsistent with official OpenAI documentation, which could lead to user confusion regarding the agent's verified capabilities.
Audit Metadata