Gemini Code Assist gone quiet on your Swift files? Work through sign-in, extension conflicts, file-type detection, and settings — the real causes of missing inline completions — and get suggestions back.
The problem: you are typing Swift in VS Code and Gemini Code Assist offers no inline completions. No ghost text, no suggestions.
Before digging deep, run a fast triage. Is the Gemini icon showing a signed-in, ready state? Do completions work in a different language like Python or JavaScript?
If completions work elsewhere but not in Swift, the issue is likely file-type detection or a Swift-specific setting. If they fail everywhere, the cause is sign-in, the extension, or connectivity.
This narrowing tells you which section below to focus on first. Most missing-completion cases come down to one of four causes: authentication, an inactive extension, wrong file type, or a disabled inline-suggestion setting.
It also helps to note exactly when the completions stopped. If they vanished right after a VS Code update, an extension update, or joining a new network, that timing is a strong clue about which fix will land. A little observation up front saves you from working through steps that do not apply.
The most common cause is a dropped or incomplete sign-in. Gemini completions require an authenticated session.
Click the Gemini icon in the activity or status bar. If it prompts you to sign in, complete the browser authentication flow with your Google account.
Sessions can expire or fail silently after a network change or a long idle period. If the panel shows an error or a stuck 'connecting' state, sign out fully and sign back in.
After re-authenticating, open a Swift file and type to test. Restored sign-in fixes a large share of 'no completions' reports on its own.
Open the Extensions panel and find Gemini Code Assist. Make sure it is installed, enabled, and not disabled for the current workspace.
VS Code lets you disable extensions per workspace. If you enabled it globally but disabled it for this project, completions will silently stop. Re-enable it for the workspace.
Check for updates. An outdated extension can break against server-side changes; update it and reload the window.
If it still misbehaves, uninstall the extension, reload VS Code, and reinstall it fresh. A clean reinstall clears corrupted local state that toggling cannot.
Gemini keys its suggestions off the detected language of the file. If VS Code does not recognize the file as Swift, completions can be weak or absent.
Look at the language indicator in the bottom-right status bar. It should say 'Swift.' If it says 'Plain Text' or something else, click it and choose Swift.
Installing the official Swift extension improves detection and gives the file a proper language mode. This alone fixes many Swift-specific completion gaps.
Also confirm the file has a `.swift` extension and is saved to disk. Unsaved, untitled buffers sometimes get inconsistent language handling.
VS Code has a global toggle for inline suggestions, and if it is off, no AI extension can show ghost text.
Open Settings (Cmd+,) and search for 'inline suggest.' Ensure 'Editor: Inline Suggest: Enabled' is turned on.
While there, search Gemini's own settings. Confirm inline completions are enabled in the extension configuration and not restricted to certain languages.
Some keyboard shortcuts or other AI extensions can hijack the Tab/accept behavior. If suggestions appear but you cannot accept them, that is a keybinding conflict, not a missing-completion problem — remap the accept shortcut.
Running multiple AI completion extensions at once — Gemini alongside another assistant — frequently causes one to suppress the other.
They compete for the same inline-suggestion slot and keybindings. The result is flickering, missing, or inconsistent completions.
Disable the other AI extensions temporarily and test Gemini alone. If completions return, you have found the conflict.
Decide which assistant owns inline completions and disable the rest, or configure them so only one provides inline suggestions while another handles chat. Do not run two inline-completion engines simultaneously.
Gemini Code Assist is cloud-backed. It needs network access to Google's services to generate completions.
If you are behind a corporate firewall, VPN, or proxy, requests may be blocked. Confirm you have general internet access, then check whether your network filters the relevant Google endpoints.
Configure VS Code's proxy settings if your environment requires one. Ask your network admin which domains to allow if completions fail only on a work network.
A quick test: try the same setup on a different network, like a personal hotspot. If completions work there, the original network is blocking Gemini's traffic.
If nothing above works, do a clean reset. Sign out of Gemini, uninstall the extension, restart VS Code completely, reinstall, and sign in again.
Check the extension's output logs. In VS Code, open the Output panel and select the Gemini channel to see error messages that reveal auth or network failures.
Confirm your account tier is active and within any usage limits. On paid Google Cloud tiers, an expired license or a project without the right API enabled can silently stop completions.
If the problem persists, consult Google's official Gemini Code Assist documentation and support channels. And remember: even with completions restored, you still build and ship the app in Xcode with an Apple Developer Program membership — the AI layer is only for authoring.
Once completions are back, a few habits keep them from vanishing on you mid-session.
Keep the extension updated but do so deliberately. When a new version lands, update and reload during a break rather than mid-task, so if behavior changes you can connect it to the update instead of chasing a phantom bug.
Settle on a single inline-completion assistant per workspace. If you like to experiment with different tools, use separate VS Code profiles so their keybindings and suggestion slots never collide in the same window.
Install the official Swift extension in any project where you write Swift. Reliable language detection is the quiet dependency behind good Swift completions, and having it in place avoids the most common Swift-specific gap.
Finally, if you work across office and home networks, note which one blocks Gemini. Knowing in advance that the corporate proxy filters the endpoints — and having the allowlist request already in with your admin — turns a confusing outage into a non-event. When suggestions do stop, run the same quick triage from the top: signed in, extension enabled, file detected as Swift, inline setting on, network reachable.
This usually means VS Code is not recognizing the file as Swift. Check the language indicator in the status bar, set it to Swift, and install the official Swift extension to improve detection.
Not well for inline completions. Multiple assistants compete for the same suggestion slot and keybindings, which suppresses completions. Let only one provide inline suggestions.
Yes. Gemini Code Assist is cloud-backed and requires network access to Google's services. Firewalls, VPNs, or proxies can block it, so test on a different network if completions fail.
Sign out, uninstall the extension, restart VS Code, reinstall, and sign in again. Then check the Gemini output logs for any auth or network errors that remain.