Codex

Codex CLI is OpenAI’s coding agent for the terminal. Use varlock to inject API keys at runtime so they never sit in a plain text .env or shell history.

For install, schema setup, and shared patterns, see the AI Tools overview.

Authentication: Codex supports ChatGPT sign-in (subscription) and API key auth (usage-based). Interactive ChatGPT login is fine for local work. Prefer an API key (via varlock) for headless/codex exec workflows.

Environment variables (see Codex environment variables):

• CODEX_API_KEY: API key for a single non-interactive codex exec run
• OPENAI_API_KEY: commonly used for API key login and custom model providers that set env_key

In a project

Add to .env.schema:

# @sensitive @required
OPENAI_API_KEY=op(op://api-local/openai/api-key)

# @sensitive @required
CODEX_API_KEY=op(op://api-local/openai/api-key)

Tip

You can use any secrets provider you want, but we like 1Password since it uses biometric authentication for local access. All the examples in this guide use the 1Password plugin (op()) to show more real world examples. See that page for plugin setup and auth options.

For interactive Codex, use varlock to pass the key into the one-time login command. Codex stores the resulting credentials for later sessions:

varlock run -- sh -c 'printenv OPENAI_API_KEY | codex login --with-api-key'
codex

For a single non-interactive run, inject CODEX_API_KEY directly:

varlock run -- codex exec "summarize the recent git changes"

From any directory

Personal schema at ~/.env.codex:

~/.env.codex

# @sensitive @required
OPENAI_API_KEY=op(op://api-local/openai/api-key)

# @sensitive @required
CODEX_API_KEY=op(op://api-local/openai/api-key)

varlock run -p ~/.env.codex -- sh -c 'printenv OPENAI_API_KEY | codex login --with-api-key'
codex

varlock run -p ~/.env.codex -- codex exec "summarize the recent git changes"

Tip

OpenAI recommends scoping CODEX_API_KEY to individual codex exec runs rather than exporting it job-wide in CI that runs untrusted repository code. varlock run keeps the key in the child process environment for that invocation only.