Troubleshooting

Fresh

Common issues and fixes for Gemini Code Assist.

Sign-In Issues

Sign-in keeps timing out

Add to VS Code settings.json:

{
  "cloudcode.beta.forceOobLogin": true
}

Extension shows "Not logged in"

1. Click spark icon > Sign out
2. Restart IDE completely
3. Click spark icon > Login to Google again

Connectivity Issues

Errors in Gemini Code Assist output window

Allow these domains through your firewall:

• oauth2.googleapis.com
• cloudaicompanion.googleapis.com

Also allow HTTP/2 traffic (required for gRPC communication).

Test connectivity:

grpc-health-probe -addr cloudaicompanion.googleapis.com:443 -tls

Success output (expected):

error: this server does not implement the grpc health protocol (grpc.health.v1.Health)

Failure output (connectivity problem):

timeout: failed to connect service "cloudaicompanion.googleapis.com:443" within 1s

For more detail, run before the probe:

export GRPC_GO_LOG_SEVERITY_LEVEL=info

Vim Plugin Issues

Can't accept completions in normal mode

Press i to enter insert mode first, then Tab to accept.

Esc dismisses both IDE and suggestions

Known issue in Vim mode — expected behavior difference from non-Vim. No workaround currently.

Chat Issues

Response gets cut off (large file)

1. Select only the specific code section you want help with
2. Add to your prompt: "only output the selected code"

Citation warnings don't persist across sessions

View persistent logs: View > Output > Gemini Code Assist - Citations

Code Suggestions

No suggestions appearing

1. Check status bar — spark icon should show "Active"
2. Verify your file's language is supported
3. Check if completions are disabled: Settings > Gemini Code Assist > Inline Suggestions

Too many cited code suggestions

Block cited suggestions entirely:

VS Code: Settings > Gemini Code Assist > Recitation: Max Cited Length > set to 0

IntelliJ: spark status bar > Configure Gemini > Advanced > Block selections that match external cited sources > OK

Agent Mode Issues

Daily quota exhausted

Add an API key to bypass quota limits:

{
  "geminicodeassist.geminiApiKey": "YOUR_GEMINI_OR_VERTEX_API_KEY"
}

Accepts either a Gemini API key or a Vertex AI API key.

MCP server not connecting

1. Type /mcp in agent chat — check if server appears
2. Verify installation: npx @modelcontextprotocol/server-github --help
3. Reload window after editing settings.json
4. Check required environment variables are set in your server config
5. For remote servers, verify the URL is accessible and auth headers are correct