Connect RaLHF to your AI.
Choose your AI tool, connect it to RaLHF, and check that it can see your libraries.
One address for every tool.
Use this remote MCP server URL in the connector setup for your AI tool.
https://backend.ralhf.ai/mcpFinish RaLHF sign-in in your AI tool, then ask it to call list_libraries. Only a successful tool response verifies the connection.
Let Muse set up the connector.
Paste the prompt below into a Muse chat that can create custom connectors.
- Ask Muse to show the RaLHF Connect card using the prompt below.
- Use Muse’s connection flow to sign in to RaLHF and approve access.
- Have Muse call
list_librariesto verify access.
Use Muse’s custom connector setup to connect RaLHF at https://backend.ralhf.ai/mcp. Show me the RaLHF Connect card and let Muse handle OAuth discovery, client registration, and its callback. After I sign in, call list_libraries to verify access. If this chat cannot create connectors, tell me that specifically; do not try to register a client through web browsing or claim RaLHF is connected.If that Muse chat cannot create connectors, it should tell you specifically. It should not register a client through web browsing or claim RaLHF is connected.
Add RaLHF in Claude.
Use Claude’s custom remote connector settings to add RaLHF to your account.
- Open Customize → Connectors, choose + → Add custom connector, name it RaLHF, and enter
https://backend.ralhf.ai/mcp. Select Add. - Find RaLHF in your connectors, select Connect, and complete RaLHF sign-in. In a conversation, use + → Connectors to enable RaLHF.
- Ask Claude to call
list_libraries. Check that the tool returns your libraries.
On Team and Enterprise plans, an owner must first add RaLHF under Organization settings → Connectors. Members can then select Connect for their own accounts. Claude’s connector instructions
Check the connection in Claude
Call RaLHF’s list_libraries tool and show me the libraries it returns. If the tool is unavailable or the call fails, tell me RaLHF is not verified as connected.Choose the OpenAI setup you use.
ChatGPT web and local Codex have different connector settings.
ChatGPT on the web
- In ChatGPT Settings, open Security and login and turn on Developer mode.
- Open Plugins → +, create a connection named RaLHF, and enter
https://backend.ralhf.ai/mcp. Complete RaLHF authorization and review the tools ChatGPT discovers. - In a new chat, add the RaLHF connection from the tools menu and ask it to call
list_libraries.
Developer mode and plugin creation depend on your account and workspace settings. If you do not see them, this ChatGPT setup is unavailable for your account. OpenAI’s ChatGPT instructions
Check the connection in ChatGPT
Use the RaLHF connection in this chat to call list_libraries and show me the libraries it returns. If the tool is unavailable or the call fails, tell me RaLHF is not verified as connected.Codex via desktop app or CLI
- In the ChatGPT desktop app, open Settings → MCP servers → Add server and name it RaLHF. Choose Streamable HTTP as the transport, enter
https://backend.ralhf.ai/mcpas the URL, then save and restart. This configuration is shared with local Codex clients. Or use the CLI commands below. - Select Authenticate when prompted and complete RaLHF sign-in. In Codex, use
/mcpto check that the server is available. - Ask Codex to call
list_librariesand confirm the tool succeeds.
Codex CLI alternative
codex mcp add ralhf --url https://backend.ralhf.ai/mcp
codex mcp login ralhf
codex mcp listChatGPT web does not read local Codex MCP settings. Set up each product where you plan to use it. Codex MCP instructions
Give RaLHF your context.
Open RaLHF on the web or desktop to add files and connect supported apps. Then your connected AI can work with the context in your RaLHF library.
Using another AI tool?
If your AI supports custom remote MCP connectors, use its own connector setup with the RaLHF URL above. Ask it to verify access with list_libraries.
Connect RaLHF as a custom remote MCP connector at https://backend.ralhf.ai/mcp.
Use RaLHF’s OAuth authorization code flow with PKCE. Discover its settings at https://backend.ralhf.ai/.well-known/oauth-authorization-server. If you can register an OAuth client automatically, use the advertised registration endpoint with your actual callback URL as the redirect URI, save the returned Client ID, then open the RaLHF sign-in page for me.
After I approve access, test the connection by calling list_libraries. Tell me it is connected only if that call succeeds. If your connector setup cannot register a client automatically, say so clearly and show me the callback URL and the Client ID field you need.