xtekky/gpt4free · error · RuntimeError

No project information found in API response.

Error message

No project information found in API response.

What it means

Project discovery loaded a response from the Code Assist load/onboard endpoint that contained no cloudaicompanionProject field, and the fallback call to onboard_managed_project also returned None (it gives up after its 10 attempts). Without a project ID the provider cannot route generate requests.

Source

Thrown at g4f/Provider/needs_auth/GeminiCLI.py:549

        try:
            load_response = await self.auth_manager.call_endpoint(
                "loadCodeAssist",
                {
                    "cloudaicompanionProject": "default-project",
                    "metadata": {"duetProject": "default-project"},
                },
            )
            project = load_response.get("cloudaicompanionProject")
            if project:
                self._project_id = project
                return project
            project = await self.onboard_managed_project(
                access_token=self.auth_manager.get_access_token(), tier_id="free-tier"
            )
            if project:
                self._project_id = project
                return project
            raise RuntimeError("No project information found in API response.")
        except MissingAuthError:
            raise
        except Exception as e:
            debug.error(f"Failed to discover project ID: {e}")
            raise RuntimeError(
                "Could not discover project ID. Ensure authentication or set GEMINI_PROJECT_ID."
            )

    async def onboard_managed_project(
        self,
        access_token: str,
        tier_id: str,
        project_id: Optional[str] = "default-project",
        attempts: int = 10,
        delay_ms: int = 5000,
    ) -> Optional[str]:
        """
        Onboard a managed project for the user, optionally retrying until completion.

View on GitHub (pinned to 973504e177)

Solutions

1. Set GEMINI_PROJECT_ID to a valid Google Cloud project that has Gemini Code Assist enabled, bypassing discovery
2. Manually trigger onboarding once in the Gemini CLI or Cloud Console with the same account, then retry
3. Re-login with an account that is eligible for Gemini Code Assist (see the 403 MissingAuthError path in onboard_managed_project)
4. Update g4f; discovery and onboarding payloads track the internal API

Example fix

// before
await provider.discover_project_id()  # raises

// after
os.environ['GEMINI_PROJECT_ID'] = 'my-gcp-project'  # skip discovery

import os

if not os.environ.get("GEMINI_PROJECT_ID"):
    # discovery will run and may fail; set it explicitly when you can
    print("warning: GEMINI_PROJECT_ID unset; project discovery will be attempted")

try:
    project = await provider.discover_project_id()
except RuntimeError as e:
    if "No project information" in str(e) or "discover project ID" in str(e):
        project = os.environ.get("GEMINI_PROJECT_ID")
        if not project:
            raise

When it happens

Trigger: discover_project_id succeeds at HTTP level but load_response.get('cloudaicompanionProject') is falsy, and onboard_managed_project finishes all attempts without a done operation, so the RuntimeError fires.

Common situations: Google account has no Cloud AI Companion / Code Assist project and auto-onboarding is blocked; region or workspace restrictions preventing managed-project creation; the free-tier onboarding endpoint changed.

Related errors

• Could not discover project ID. Ensure authentication or set
• Project ID discovery failed - set ANTIGRAVITY_PROJECT_ID in
• Invalid thinking mode: {think_value!r}
• Thinking mode must be an integer between 0 and 4
• No refresh token found in GCP_SERVICE_ACCOUNT.

AI-assisted analysis of xtekky/gpt4free@973504e177 (2026-08-14). Data as JSON: /api/errors/5a3a96c49e9c2224. Report an issue: GitHub.