A new open-source tool, Gemini Config Checker, helps users identify proxy configurations that can access Google AI Studio and Gemini in restricted regions. The application, currently in early pre-release (version 0.1.0), tests proxies by simulating how a real user would access the services — through a signed-in browser session — rather than relying on simple connectivity checks that may give false positives.
The tool operates in two modes: Simple and Advanced. In Simple mode, it evaluates a list of user-provided proxy configurations (via links, subscription URLs, or base64) to determine which ones successfully reach Google’s AI services. Advanced mode allows chaining free configurations behind a working ‘base’ proxy — useful when the base connects but its exit location is blocked by Google. This is particularly relevant for users relying on Cloudflare-based tunnels that may route through geolocations like the US, Hong Kong, or Singapore, where access has reportedly been denied.
How the tool verifies access
Unlike basic ping or HTTP checks, which return positive results even in blocked regions, Gemini Config Checker opens actual browser windows (Chrome, Edge, or Brave) in an isolated profile. It signs into Google and observes whether the service redirects to /docs/available-regions (for AI Studio) or displays the message “Gemini isn’t currently supported in your country”. This method captures Google’s real-time, account-specific region enforcement, which applies after page load and cannot be detected via API calls alone.
Each test takes 20–40 seconds per configuration. The application supports Windows, macOS, Linux, and Android, though only Windows has been fully tested end-to-end. On Android, the app runs in connectivity-only mode due to platform restrictions on driving signed-in browser sessions.
Technical implementation and dependencies
The tool embeds XTLS/Xray-core as its proxy engine and uses Wails for the desktop interface. It applies anti-DPI (Deep Packet Inspection) techniques — such as TLS ClientHello fragmentation (“finalmask”) and custom cipher suites — based on a recipe from the PattN client by patterniha. These settings help evade detection in restrictive networks, such as those in Iran, where standard Cloudflare configurations often fail.
To support chaining, the app runs a local forwarder that replicates PattN’s fragmenting logic, since the standard Xray-core only supports basic fragmentation. Users can import configurations with embedded parameters (finalmask, ALPN, ciphers) or apply default anti-DPI profiles manually.
Export and compatibility
Results can be exported as shareable links, base64 strings, or full Xray JSON configurations. The “Copy with chain” function generates a ready-to-use chained proxy config for clients like v2rayN, Hiddify, or PattN. Exports are tailored: configurations for PattN include finalmask settings, while those for standard Xray/v2rayN omit them to ensure compatibility.
The tool pulls free config lists from 0xRadikal/Free-v2ray-Configs and supports chaining behind base proxies created via bia-pain-bache/BPB-Worker-Panel. It also integrates with MatinSenPai/SenPaiScanner for identifying clean Cloudflare IPs.
What it means
For users in region-restricted markets, this tool offers a practical way to validate access to Google’s AI services without relying on unverified public proxies. However, results are not guaranteed long-term: the exit IP of a Cloudflare Worker can change between connections, so a working config today may fail tomorrow. Businesses depending on stable access should treat results as temporary and retest regularly.
The reliance on signed-in Google accounts also introduces risk: using a primary account may trigger suspicious login alerts due to traffic from multiple IPs. The developers recommend using a secondary account.
For IT teams in the UK, UAE, US, and Kazakhstan, this highlights the fragility of depending on consumer-grade proxy networks for accessing critical AI tools. While the tool itself is free and open-source, it reinforces technical debt through dependency on third-party configs and Cloudflare infrastructure, with no service-level assurances.
Organisations should assess whether such methods align with data governance policies, especially when handling sensitive inputs in AI workflows. The lack of official support or audit trails makes this unsuitable for regulated environments.
The tool is available via GitHub releases with SHA256 checksums. Builds are unsigned or signed with throwaway keys, requiring manual trust decisions on macOS and Windows. The web server version allows deployment on remote machines via SSH tunneling to avoid exposing the UI.
Despite its utility, the project remains in early development. macOS, Linux, and Android builds are untested on real devices, and users are encouraged to report findings. No official response from Google has been reported regarding the tool or its implications for regional access enforcement.