HKUDS/Vibe-Trading · error · RuntimeError

OpenAI Codex OAuth requires oauth-cli-kit. Install dependenc

Error message

OpenAI Codex OAuth requires oauth-cli-kit. Install dependencies, then run: {_CODEX_LOGIN_COMMAND}

What it means

_load_codex_oauth_provider imports OPENAI_CODEX_PROVIDER from oauth_cli_kit; failure means the OAuth provider constants for ChatGPT/Codex login are unavailable, and login or token refresh cannot proceed.

Source

Thrown at agent/src/providers/openai_codex.py:204

    """Return the access token's real JWT expiry, with stored expiry fallback."""
    access = getattr(token, "access", "")
    claims = _decode_jwt_claims(access) if isinstance(access, str) else {}
    jwt_expiry = claims.get("exp")
    if isinstance(jwt_expiry, (int, float)) and not isinstance(jwt_expiry, bool):
        return int(jwt_expiry * 1000)
    try:
        stored_expiry = int(getattr(token, "expires"))
    except (TypeError, ValueError):
        return 0
    return stored_expiry * 1000 if stored_expiry < 100_000_000_000 else stored_expiry


def _load_codex_oauth_provider() -> Any:
    """Load oauth-cli-kit's provider constants without using its token cache."""
    try:
        from oauth_cli_kit import OPENAI_CODEX_PROVIDER
    except ImportError as exc:
        raise RuntimeError(
            f"OpenAI Codex OAuth requires oauth-cli-kit. Install dependencies, then run: {_CODEX_LOGIN_COMMAND}"
        ) from exc
    return OPENAI_CODEX_PROVIDER


@dataclass
class CodexToolCall:
    """Internal tool-call representation compatible with ChatLLM parsing."""

    id: str
    name: str
    arguments: dict[str, Any]

    def as_langchain_tool_call(self) -> dict[str, Any]:
        return {"id": self.id, "name": self.name, "args": self.arguments}


@dataclass

View on GitHub (pinned to 80ffdda44c)

Solutions

Install oauth-cli-kit, then re-run the login command shown in the message (_CODEX_LOGIN_COMMAND)
Verify: python -c "from oauth_cli_kit import OPENAI_CODEX_PROVIDER"
Ensure you installed into the same interpreter the agent runs under (pip -m or match the venv)

Example fix

# before
RuntimeError: OpenAI Codex OAuth requires oauth-cli-kit...

# after
pip install oauth-cli-kit
vibe-trading provider login openai-codex  # re-run login

Defensive patterns

Strategy: try-catch

Validate before calling

try:
    from oauth_cli_kit import OPENAI_CODEX_PROVIDER  # noqa: F401
except ImportError:
    raise SystemExit('pip install oauth-cli-kit then re-run login')

Try / catch

try:
    provider = _load_codex_oauth_provider()
except RuntimeError as e:
    if 'oauth-cli-kit' in str(e):
        subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'oauth-cli-kit'])
        provider = _load_codex_oauth_provider()
    else:
        raise

Prevention

Bundle oauth-cli-kit wherever Codex login is offered
Verify the active interpreter matches the install target

When it happens

Trigger: Invoking login_openai_codex or _refresh_codex_token when oauth-cli-kit is missing; the message also embeds the recommended login command to re-run after installing.

Common situations: Same as other oauth-cli-kit errors: dependency not installed in the active environment; stale venv after switching branches that changed extras.

Understand the failure class

Background: "X is not installed. Please install it with pip install Y": missing optional dependency errors — ImportError/ValueError raised when a library's optional extra was never installed — this error's family across 22 libraries.

Related errors

AI-assisted analysis of HKUDS/Vibe-Trading@80ffdda44c (2026-08-28). Data as JSON: /api/errors/b3e43c5ae139b61f. Report an issue: GitHub.