ProxyPal: Run Claude, ChatGPT and Copilot Subscriptions Through One Local Proxy
A desktop app that lets you use your AI subscriptions (Claude, ChatGPT, Gemini, GitHub Copilot) with any coding tool. Wraps CLIProxyAPI with a clean UI for managing connections and tracking usage.
At a glance
- What is it?
- ProxyPal is a Tauri desktop app that wraps CLIProxyAPI so coding tools can talk to your existing AI subscriptions over an OpenAI-compatible endpoint. It is a convenience layer with a real dependency underneath, and that dependency is the thing to evaluate.
- Who is it for?
- Adopt ProxyPal if you already pay for one or more AI subscriptions, want a GUI for connecting accounts and reading usage, and are comfortable that the actual proxying is done by CLIProxyAPI rather than by ProxyPal itself. Skip it if you need a signed macOS build, a headless server deployment, or a proxy whose routing logic you can audit inside this repository, because the proxying layer lives in a separate project.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 3 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The subscription mismatch ProxyPal is built around
Most coding assistants assume you will paste an API key and pay per token. Many developers instead hold a Claude, ChatGPT, Gemini or GitHub Copilot subscription and want those same models inside Cursor, Cline, Continue or Claude Code. Those clients speak the OpenAI-compatible chat completions shape; the subscriptions do not expose one directly. ProxyPal's stated purpose is to close that gap: the README asks, "You're paying for Claude, ChatGPT, or GitHub Copilot. Why can't you use them in your favorite coding tool?"
The intended audience is individual developers on desktop machines, not platform teams. The supported platform table lists macOS on Apple Silicon and Intel, Windows x64, and Linux x64 as a .deb package. There is no documented server build, no Docker image, and no headless mode, so anyone hoping to run this on a shared build box is outside the design.
What ProxyPal actually does versus what CLIProxyAPI does
This is the distinction that matters most when judging the project. ProxyPal describes itself as a "Native desktop app wrapping CLIProxyAPI." The proxying, account handling and model routing are performed by CLIProxyAPI, a separate repository by a different author (router-for-me). ProxyPal supplies the desktop shell: account connection flows, a dashboard, usage analytics, request monitoring, and auto-configuration of installed CLI agents.
The stack is split accordingly. The frontend is SolidJS with TypeScript and Tailwind; the backend is Rust on Tauri v2; CLIProxyAPI sits underneath as the proxy. The repository layout reflects that, with src/ for the frontend, src-tauri/ for the Rust side, and a cli-proxy-api-management/ directory at the top level. The README's contributor notes point at src-tauri/src/lib.rs for agent detection logic, which is where the auto-configure behaviour lives.
The practical consequence: if a provider changes its authentication flow or a model stops responding, the fix may land in CLIProxyAPI before it lands here. ProxyPal's release cadence is its own, and the two are not versioned together in anything the README shows.
Installing ProxyPal and pointing a coding tool at it
There is no package manager install. The README's Quick Start begins with downloading from the Releases page, launching the app, starting the proxy, and connecting accounts via OAuth or auth files. Only after that do you point a client at the local endpoint. The port is fixed in the documentation as 8317.
The macOS build is not signed with an Apple Developer certificate. The README gives this workaround for the case where Gatekeeper blocks the app:
xattr -cr /Applications/ProxyPal.appRunning that clears the extended attributes on the app bundle. Expect to run it once after dragging ProxyPal into Applications. On Windows x64 and Linux x64 (.deb) the README does not describe an equivalent step.
Once the proxy is running and at least one account is connected, the client side is a base URL change. The README states the endpoint as http://localhost:8317/v1, which is the value you enter wherever your tool asks for an OpenAI-compatible base URL. The README does not give a shell command for this, so the exact configuration key depends on your client.
If you want to build from source rather than use a release, the README documents the Tauri workflow and its checks:
pnpm install
pnpm tauri devpnpm check:parallel # check:ts + lint + format:check (parallel)
cd src-tauri && cargo checkBuilding from source also means supplying Google OAuth credentials, because .env.example defines ANTIGRAVITY_CLIENT_ID and ANTIGRAVITY_CLIENT_SECRET and notes they are injected at compile time via env!() in quota.rs. Those are for Antigravity token refresh. Without them, that path is not configured.
Where ProxyPal is the wrong tool
The unsigned macOS build is the most concrete limitation, and it is acknowledged rather than fixed. The README's own workaround tells you the project has not yet obtained an Apple Developer certificate. For a managed fleet, that alone can be disqualifying, since it means either disabling Gatekeeper protections or running a per-machine command.
The second limitation is architectural. ProxyPal is a GUI wrapper, and the README documents no command-line interface for the app itself, no configuration file format, and no way to run it without a desktop session. If your workflow is a remote dev container, a CI runner, or an SSH session, this is not the shape you want. The README also does not document rollback, so if a version update changes behaviour you have no described path back other than reinstalling an earlier release artifact.
Third, the trust question is real and the README does not address it. A local proxy that holds OAuth credentials for your AI accounts sits between your editor and the provider. The repository is MIT licensed and the source is available, which is the strongest answer available here, but the README contains no security section, no threat model, and no statement about where credentials are stored on disk. If that matters to your environment, read the source in src-tauri/ before connecting an account.
ProxyPal compared with running CLIProxyAPI directly
The obvious alternative is the thing ProxyPal wraps. CLIProxyAPI is a standalone project; running it yourself gives you the proxy without the desktop shell. The difference in approach is not cosmetic. With CLIProxyAPI alone you configure accounts through its own mechanism, you get no dashboard, no usage charts, no request log view, and no automatic detection of installed agents. You also get no Tauri app to install, no Gatekeeper workaround, and no dependency on ProxyPal's release schedule.
ProxyPal's value sits entirely in that shell. The README lists usage analytics covering requests, tokens, success rates and estimated savings, request monitoring with response times and status codes, and auto-configure for detected CLI agents. If you would otherwise be editing config files by hand for each client, that is the trade. If you are comfortable in a terminal and only need one client wired up, the wrapper adds a layer you will have to keep updated without giving you much back.
A second alternative is simply using each provider's official API keys in your coding tool and paying per token. That avoids the credential-holding proxy entirely, at the cost of the subscription you already bought going unused for this purpose.
Maintenance cost and licence terms
ProxyPal is not archived, and the last push to the default branch was on 2026-09-13. Releases are frequent: v0.4.53 on 2026-09-09, v0.4.52 on 2026-08-21, and v0.4.51 on 2026-08-18, while package.json carries version 0.4.54. That pace cuts both ways. You get fixes quickly, and you also get a moving target, with no documented rollback path in the README.
Upgrade cost is mostly the desktop app itself, since Tauri bundles the updater plugin and the release artifacts are the distribution channel. The hidden cost is the sidecar. The repository includes scripts/update-sidecar.mjs and a test:sidecar script, which indicates the bundled CLIProxyAPI binary is refreshed by a script rather than by the package manager. When CLIProxyAPI changes, that script is how the change reaches ProxyPal.
The project is MIT licensed, which permits commercial and private use and modification. The README does not include a NOTICE file or document third-party licence obligations for the bundled CLIProxyAPI binary, so if you redistribute a build, check the upstream project's licence yourself. Nothing here is legal advice.
Editorial conclusion
Adopt ProxyPal if you already pay for one or more AI subscriptions, want a GUI for connecting accounts and reading usage, and are comfortable that the actual proxying is done by CLIProxyAPI rather than by ProxyPal itself. Skip it if you need a signed macOS build, a headless server deployment, or a proxy whose routing logic you can audit inside this repository, because the proxying layer lives in a separate project. Before committing, verify the release artifact for your platform on the Releases page, confirm your coding tool accepts a custom OpenAI-compatible base URL, and read CLIProxyAPI's own documentation for the request path, since ProxyPal's README does not describe how it forwards or transforms traffic.
Frequently asked questions
What is ProxyPal?
It is a native desktop app that wraps CLIProxyAPI so you can use AI subscriptions such as Claude, ChatGPT, Gemini and GitHub Copilot with coding tools like Cursor, Cline, Continue and Claude Code. It provides the UI for connecting accounts and tracking usage, while CLIProxyAPI performs the actual proxying.
Is ProxyPal safe to use?
The repository is MIT licensed and the source is public, including the Rust backend in src-tauri/, so the code can be reviewed. The README does not document where credentials are stored on disk or include a security section, so anyone connecting an account should read the source first. Note that the macOS build is not signed with an Apple Developer certificate.
What port does ProxyPal listen on?
The README gives the endpoint as http://localhost:8317/v1, and that is the base URL you point an OpenAI-compatible client at. The port is documented as part of the Quick Start rather than as a configurable setting.
What is a ProxyPal alternative?
Running CLIProxyAPI directly is the closest alternative, since ProxyPal wraps it. You lose the dashboard, usage analytics, request monitoring and agent auto-configuration, but you avoid the desktop app, the unsigned macOS build and ProxyPal's own release cycle.
Does ProxyPal work on Windows and Linux?
The supported platform table lists Windows x64 and Linux x64 as a .deb package alongside macOS on Apple Silicon and Intel. The README's Gatekeeper workaround is macOS-specific and no equivalent step is documented for the other platforms.
Community notes