Everything on this site is also an MCP server, so an agent can read it directly instead of scraping these pages. It is read-only and needs no account, no key and no sign-in.
https://unbias.me/mcp
Custom connectors need a paid Claude plan. Open the app, click the menu icon at the top left, hover over File and choose Settings. Click Connectors in the sidebar, then Add at the top right, then Add custom connector. Paste the endpoint above and click Add. No sign-in step follows because the server needs none.
Connectors belong to your Claude account, so the same connector shows up on claude.ai in the browser. There, Settings is under your profile icon.
In a chat, click the plus at the bottom left of the message box and hover over Connectors to turn it on. Ask Claude to call the tags tool first so it knows what the tags mean.
claude mcp add --transport http unbias https://unbias.me/mcp
In a terminal:
codex mcp add unbias --url https://unbias.me/mcp
Or in the ChatGPT desktop app, open Codex, then Settings, then MCP servers. Click Add server, give it a name, choose Streamable HTTP and paste the endpoint above. Save, then click Restart. The desktop app, the Codex command line and the IDE extension share one configuration, so adding it in one place adds it everywhere.
Custom servers in ChatGPT chats need a paid plan and developer mode, which OpenAI offers on the web. Open Settings, then Security and login, and turn on Developer mode. Go to the apps page and click the plus button to create an app for a remote MCP server. Paste the endpoint above and choose No Authentication. In a chat, open the plus menu, choose Developer mode and pick the app.
Paste this to an agent that can edit its own configuration:
Add the unbias.me MCP server at https://unbias.me/mcp over HTTP transport. It is read-only, needs no authentication, and gives you news links with their measured political bias. Once it is connected, call its tags tool first so you know what the tags mean.
Each link comes back with its rewritten headline, the original, its topics, and the full probability distribution behind every tag rather than just the label. Only articles we could read in full are listed, so every link carries scores.