Skip to content

README.md

Latest commit

 

History

History
170 lines (137 loc) · 8.87 KB

README.md

File metadata and controls

170 lines (137 loc) · 8.87 KB
Raw

Relmio gateway android logo

Relmio

Relmio helps you test a supported ChatGPT sign-in in a hosted chat, connect self-hosted n8n through a separate private Docker sidecar, or install a loopback-only local endpoint. It keeps ChatGPT/Codex credentials, OpenAI Platform API keys, and local client capabilities as three different contracts.

Quick install

On macOS, Linux, WSL, or Git Bash:

npx --yes --ignore-scripts relmio@latest

The local wizard prints a private 127.0.0.1 URL, checks Docker, presents the exact plan, and asks for final confirmation before it writes local files or a VPS sidecar. Other install options are on the hosted install page.

Endpoints

Option Contract Intended credential/client
OpenAI API gateway OpenAI-compatible /v1 User-supplied Platform API key; local apps and configured browser origins
Codex App Server Experimental JSON-RPC over WebSocket ChatGPT sign-in plus a high-trust local capability; trusted native clients
Codex Chat Adapter Experimental POST /chat (JSON or opt-in SSE) ChatGPT sign-in plus a local bearer; trusted local backends or development servers
n8n sidecar Private http://n8n-openai-oauth:10531/v1 Existing n8n Docker network only
n8n AI Assistant companion Private sandbox; opt-in SearXNG web search User-owned OpenAI Platform API key entered directly in n8n

The Codex targets are not generic /v1 endpoints. ChatGPT sign-in is never an OpenAI Platform API key or authorization for arbitrary OpenAI API calls.

n8n AI Assistant companion

Run relmio assistant to open the dedicated local wizard for the self-hosted AI Assistant sandbox and an explicit, opt-in SearXNG web-search companion. Web search is off by default and its exact boolean selection is bound to the reviewed plan and confirmation. It reuses the verified SSH, read-only discovery, and selected existing Docker network flow, but never reads ChatGPT/Codex OAuth. ChatGPT/Codex subscription sign-in is not an OpenAI Platform API key and Relmio does not present it as the compliant model route.

AI Assistant is Preview: review every generated workflow before use. The companion uses n8n's self-hosted privileged Docker-in-Docker runner for advanced/local testing and keeps every host port unpublished. n8n recommends Daytona for production sandboxing. Enter a user-owned Platform API key directly in n8n; Relmio never handles it. Keep your current supported model selection, including openai/gpt-5.6-sol while n8n continues to accept it.

The result view also shows an optional custom OpenAI-compatible route for a separately deployed, Platform-key-backed Relmio endpoint. Its private Relmio client credential is not an OpenAI-issued API key. The existing OAuth sidecar remains experimental/private/policy-uncertain and is never auto-selected or described as policy-approved. See the AI Assistant guide.

The experimental Chat Adapter's SSE stream succeeds only after its terminal: completed event. Its local tester stays behind the setup-token- protected wizard and uses a short-lived encrypted credential handoff, never a direct browser-to-adapter request.

OpenAI policy evidence and limits

Relmio's supported paths follow published OpenAI product boundaries rather than treating a ChatGPT subscription as a reusable API key:

  • Relmio's maintainer was accepted into the Codex for Open Source program for this project in August 2026 and received the program's limited-duration ChatGPT Pro benefit. The private acceptance email is retained by the maintainer and is not published because it contains personal account information.
  • OpenAI's advanced Codex configuration documents custom model providers and OSS mode with local providers such as Ollama and LM Studio.
  • Thibault “Tibo” Sottiaux, Codex Lead at OpenAI, has publicly stated that the Codex App, CLI, and SDK can use open-source models, and separately distinguished supported Sign in with ChatGPT clients from unsupported subscription-to-API conversion, resale, or multi-user sharing (open-model statement, account-use statement).
  • OpenAI CEO Sam Altman publicly announced ChatGPT-account sign-in for OpenClaw (statement).

These sources are evidence for the specific documented patterns, not a blanket OpenAI endorsement, legal opinion, contractual amendment, or promise that every third-party adapter or future release is compliant. Program acceptance supports the maintainer and open-source work; it is not a protocol-by-protocol compliance certification. Relmio therefore keeps native Codex traffic in the official App Server lifecycle, does not present ChatGPT credentials as a generic /v1 API key, and requires a user-owned OpenAI Platform API key for the n8n AI Assistant model route. Relmio prohibits account pooling or sharing, credential forwarding, subscription-to-API conversion or resale, and rate-limit or safeguard bypass. The legacy n8n OAuth sidecar remains experimental/private/policy-uncertain. See the complete policy evidence and scope.

Critical security boundaries

  • Local endpoints bind only to 127.0.0.1; do not expose them through a LAN, reverse proxy, domain, or public IP.
  • The n8n installer creates a separate sidecar. It never edits the existing n8n Compose file or image and never publishes port 10531 on the VPS host.
  • The Chat Adapter rejects browser origins. Its new in-wizard tester calls only the setup-token-protected local wizard, never the adapter from the browser.
  • Local capabilities and the Chat Adapter bearer are sensitive. Do not put them in browser code, logs, or a shell command line.

ChatGPT sign-in lifetime

ChatGPT/Codex sign-in tokens expire, but the official Codex client refreshes them automatically during active use before they expire, so active sessions usually continue without another browser login. The official OpenAI authentication documentation does not publish a fixed 10-day lifetime; do not plan around one. This provider credential is separate from Relmio's local capability, which remains valid until you rotate it.

Common problems

  1. Docker is not running. Start Docker Desktop or Docker Engine with Compose, then reopen one fresh wizard session. See Docker troubleshooting.
  2. Authentication fails. Close old wizard and sign-in tabs, use the newest printed wizard URL, and retry the device-code flow. See authentication troubleshooting.
  3. Local image build failed. The wizard intentionally hides Docker stderr and local paths. Check Docker, disk space, and registry connectivity, then start a fresh reviewed plan. See local image build troubleshooting.

Documentation

For source-level guides, see docs/. Security guidance is in docs/security.md. Relmio is licensed under the Apache License 2.0.

Legal

Relmio is an unofficial, community-maintained project. It is not affiliated with, endorsed by, or sponsored by OpenAI.

Treat ChatGPT/Codex credentials like passwords. Use only your own account, keep credentials private, and never pool, share, or redistribute access tokens. Any request made with those credentials must be authorized by the account owner.

You are responsible for following OpenAI's Terms of Use, Usage Policies, and any other agreement that applies to your account.

Warning

Do not bypass rate limits, restrictions, or safeguards.

Relmio is provided as-is without warranties. OpenAI or an upstream service may change or discontinue access at any time, and you accept the risks of using this experimental community project.