codex-chatgpt-web

   

macOS Intel · All releases

English · 简体中文 · 日本語 · 한국어

Get started · What’s new · Architecture · Troubleshooting

Use the ChatGPT Web models available on your account, including Pro, from Codex’s native model picker—with ChatGPT Web’s separate usage limits, without spending your Work or Codex quota. Keep the same interface, tasks, images, and streaming. Full harness mode connects ChatGPT to the current task’s files, terminal, tools, and approvals through MCP. Conversations stay tied to your Codex task, so you can keep working as the context grows. ## Get started **Available models:** Free/Go → **Luna / Think**. Accounts with reasoning controls → **Instant–High**, plus **Extra High** and **Pro** when available. The launcher detects what your account can use. 1. **Install the launcher** using the download for your system above. 2. **Sign in to ChatGPT** in the embedded browser and run the browser smoke test. 3. **Install models**, restart Codex once, and choose a **ChatGPT Web — …** model. 4. **For coding with tools**, open **MCP** in the launcher and complete the Full harness setup below. The app includes its browser and runtime. No separate Chrome, Node, or Bun installation is needed. Terminal install, updates & repair Quit the launcher before updating. These installers select the platform and architecture, verify the published checksums, and preserve your ChatGPT profile and launcher settings. **macOS / Linux** ```bash curl -fsSL https://github.com/miuuyy/codex-chatgpt-web/releases/latest/download/install-launcher.sh | sh ``` **Windows PowerShell** ```powershell irm https://github.com/miuuyy/codex-chatgpt-web/releases/latest/download/install-launcher.ps1 | iex ``` Models, modes & MCP setup Automatic modes offer Luna/Think when the account has no reasoning selector; otherwise Instant–High, with Extra High and Pro available independently when exposed by the account. | Mode | Sending messages | Local Codex tools | | --- | --- | --- | | **Browser-only** | Automatic | No | | **Full harness (With Automation)** | Automatic | Yes, through MCP | | **Zero Risk** | Paste and send manually | Yes, through a separate MCP connector | Zero Risk does not read or operate the ChatGPT page. Choose the model and `Codex Zero Risk` connector yourself, paste and send the prepared prompt, then confirm **Sent** in the launcher. Automatic model entries each select a fixed ChatGPT mode; Codex’s Effort and Speed rows do not override it. ### Full harness Full mode connects ChatGPT's tool calls back to the current Codex task through the official [OpenAI tunnel-client](https://github.com/openai/tunnel-client). The tunnel is outbound: it does not expose a public IP, open an inbound port, or require router forwarding. The launcher's **MCP** page guides the complete setup. For the exact clicks, see the [video walkthroughs](TROUBLESHOOTING.md). > **Limits** > > See [Limits](https://github.com/miuuyy/codex-chatgpt-web/discussions/309) for the current > ChatGPT message allowances for **GPT-5.6 Sol Pro** and **GPT-6 Astra**. Context limits depend on > the account type and selected effort. Plus Medium/High uses a measured 90,000-token window, or > up to 270,000 tokens with experimental **3× context** enabled, with native Codex compaction > supported throughout. 1. Finish the required setup, open **MCP**, create the Tunnel and regular API key, then press **Connect harness**. 2. Enable ChatGPT **Developer Mode** and create a new Tunnel connector named exactly **Codex Native2**, with **Authentication: None** and **Allow all actions**. 3. Run **Verify runtime** to confirm that **Codex Native2** is attached and available. Write/modify actions also require the ChatGPT workspace and its administrator policy to permit them. See [developer mode and MCP apps](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt). Unexpected approval prompts fail closed unless `--auto-approve-tool-calls` is explicitly enabled; that option clicks **Allow once**, never a permanent grant. Diagnostics & subagents Use **Activity** for safe local diagnostics and **Settings → Run doctor** for end-to-end health. Settings can also cancel a retained browser turn or remove the Codex integration before uninstall. Set `CODEX_CHATGPT_WEB_BROWSER_DIAGNOSTICS=1` only when every browser checkpoint needs a screenshot. New installs use **Compatibility V1** for cross-backend subagents. **Native** preserves Codex's own feature settings and enables plaintext Web-to-Web V2 delegation. Restart Codex and start a new task after changing the protocol: ```bash codex-chatgpt-web subagents status codex-chatgpt-web subagents compatibility-v1 codex-chatgpt-web subagents native ``` Requirements & security - This is unofficial browser automation, not an OpenAI API. ChatGPT UI changes can break selectors; drift fails explicitly instead of silently switching model or transport. - Browser state is a sensitive login artifact, and the loopback listener is reachable by processes running as the same local user. Never share the launcher profile; use a trusted workstation. - Release packages currently target macOS 13+ (arm64/x64), Windows x64, and Linux x64. Runtime, tests, and packaging are gated on all three in CI; account-bound browser and MCP flows use the separate [release validation](docs/release-validation.md). - Builds are not yet platform-signed, so Gatekeeper or SmartScreen may warn. The installers verify the published SHA-256 manifest before installation. Read the complete [architecture](docs/architecture.md) and [security model](docs/security-model.md) before enabling full mode. Report vulnerabilities through [SECURITY.md](SECURITY.md). Temporary Chat is a [ChatGPT privacy mode](https://help.openai.com/en/articles/8914046-temporary-chat-faq); prompts are still processed by OpenAI. Validation coverage: [release validation](docs/release-validation.md). This is independent software and is not affiliated with or endorsed by OpenAI. Use it only with your own account and in accordance with applicable [Terms of Use](https://openai.com/policies/terms-of-use/) and workspace policies; it does not bypass authentication or access controls. Run from source & develop ```bash git clone https://github.com/miuuyy/codex-chatgpt-web.git && \ cd codex-chatgpt-web && \ bun run app ``` This source path requires Bun 1.4.0. The command installs locked dependencies and opens the app. ```bash bun run app bun run dev:launcher bun run src/cli.ts dev status bun run dev:chat compaction-lab "Reply with exactly: DEV READY" bun run verify bun run smoke:subagents bun run app:package ``` `dev:launcher` uses a separate profile and account under `~/.codex-chatgpt-web-dev`. `dev:chat` exercises the real browser and compaction paths with explicit simulated tool results, without changing your normal Codex route. See the [DEV chat harness](docs/dev-chat.md) for setup and commands. ## Star History --- [Troubleshooting](TROUBLESHOOTING.md) · [Security](SECURITY.md) · [Contributing](CONTRIBUTING.md) · [MIT license](LICENSE) · [CI](https://github.com/miuuyy/codex-chatgpt-web/actions/workflows/ci.yml) Also by me: [ChatGPT Persona Voice](https://github.com/miuuyy/ChatGPT-Persona-Voice) — local, near-real-time custom voices for ChatGPT and Codex.