1Password has announced a partnership with OpenAI aimed at improving security for AI-powered software development systems. The collaboration focuses on addressing growing concerns around how AI coding agents access, handle, and protect enterprise credentials during development workflows.
As part of the partnership, the companies introduced a new integration for OpenAI Codex that allows AI coding agents to securely access credentials without exposing sensitive information in prompts, source code, repositories, terminals, or the AI model’s context window.
The development comes at a time when AI-assisted coding tools are increasingly becoming central to software development processes across the technology industry. However, the rise of agentic AI systems in programming environments has also created new cybersecurity risks related to credential exposure, misuse, and data leakage.
According to 1Password, many software development environments currently store credentials inside .env files, scripts, or repositories where they may become vulnerable to theft or accidental exposure. The company stated that AI coding agents can unintentionally increase this risk because they often require access to multiple databases, APIs, deployment systems, and development tools during application creation.
“Every action that AI coding agents take against a database, an API, or a deployment pipeline requires access to credentials,” explained Dennis Kromhout van der Meer and Robert Menke in a company blog post. “Today, these credentials typically live in .env files, scripts, or hardcoded in repositories, where they can be easily exfiltrated and are difficult to govern and audit.”
To address these concerns, 1Password introduced its new Environments MCP Server for Codex. The system provides credentials directly inside development workflows while keeping those secrets outside prompts, codebases, and model memory. Credentials are generated only when required for a task and are discarded immediately after use.
“As coding agents take on more of the software development lifecycle, the question isn’t whether to give them access, but how,” said Nancy Wang. “A credential that persists is already compromised. That’s why just-in-time credentials are the only viable security model for AI-native development.”
The system uses 1Password’s vault technology to ensure credentials remain encrypted and centrally managed. User authentication is required at the moment of access, and secrets only exist temporarily in memory for authorized processes before being removed automatically.
The partnership reflects a broader industry shift toward integrating AI agents into enterprise operations while maintaining strong security oversight. 1Password stated that coding agents are only the beginning of a future where AI systems across multiple industries will require secure, temporary access to real-world systems and infrastructure.
The company believes the integration with Codex represents an early framework for building secure access layers that allow AI agents to operate efficiently without directly controlling or permanently storing sensitive credentials.
