| name | openai-sdk-helpers |
| description | Install and use openai-sdk-helpers from Git with OpenAI credentials loaded only from a local .env file. |
OpenAI SDK Helpers
Local-only credential policy
Use only credentials stored in the user's local project .env. Never request secret values in chat, upload them, print them, or copy them into Codex, Git, plugin files, hosted configuration, URLs, or command history.
First-use setup
Install the repository version locally:
python -m pip install --upgrade "git+https://github.com/fatmambot33/openai-sdk-helpers.git"
Before any OpenAI network operation, run:
openai-helpers-credentials doctor
When the check fails, guide the user through:
openai-helpers-credentials configure
The wizard securely prompts for OPENAI_API_KEY without echoing it and optionally accepts OPENAI_PROJECT and OPENAI_ORG_ID. It writes only to the current project's .env, applies restrictive permissions where supported, refuses accidental overwrites, and warns when Git-ignore protection is missing.
Run the doctor again after setup. It validates file existence, required variable names, permissions, and .gitignore coverage without displaying credential values. Stop network operations until it passes.
Use --env-file PATH only for a deliberate non-default local file. Never fall back to hosted secrets, Codex-managed credentials, remote secret stores, or credentials supplied in prompts.
Workflow
- Run
openai-helpers-credentials doctor.
- Load
.env locally with python-dotenv.
- Use the package's Codex registry and typed helpers.
- Prefer structured outputs and validators.
- Redact credentials and authorization headers from outputs and errors.
- Run relevant tests and type checks after code changes.
from dotenv import load_dotenv
from openai_sdk_helpers import CodexPluginRegistry
load_dotenv(".env")
registry = CodexPluginRegistry()
print(registry.list_commands())
