MCP Dashboard for the Servers You Use

Connect Your MCP servers

Open Bring your MCP servers In Connect, open Your data and choose Bring your MCP servers.
Paste your settings Paste the settings file of the AI app that already has your servers, or a list of addresses. With your computer linked, the board lists the ones it found there instead.
Connect them Connect every server that needs no sign-in in one click. Click Sign in for the rest, and the service's own sign-in opens.
Pick what to show Choose one of a server's read tools for a widget, or ask your agent. You see what the widget will say before it is added.

Your agent works on your board, and everything it does carries its name and can be undone. How your agent works with the board

What to ask your agent

“List what my connected MCP servers can read” Your agent checks every connection and tells you which tools can feed a widget.

“Put my open Linear issues and unresolved Sentry errors side by side on the board” Two live widgets, each reading its own server while your board is open.

“Show my latest Vercel deployments under the issues” The agent adds the widget and places it where you asked.

What you see from Your MCP servers

Reads Claude Desktop, Claude Code, Cursor, VS Code, Windsurf (now Devin Desktop), Codex and Gemini CLI settings
Connects every server that needs no sign-in in one click
Signs in to the rest with the service's own sign-in
Adds any remote MCP server by its address
Runs only the tools that read, so nothing in your services changes
Shows what a widget will say before it is added
Finds the servers in your apps on a linked computer, keys left behind

Your MCP servers on Context Board: the full guide

You already did the hard part. Your AI app has your MCP servers set up, signed in and working. Linear, Sentry, Vercel, Stripe. Your agent can ask any of them anything.

What it cannot do is show you the answers tomorrow morning without being asked again.

What Bring your MCP servers does

Open Connect, then Your data, and choose Bring your MCP servers. Paste the settings file of the AI app that already has them, or just a list of addresses. The board reads it on the page and lists every server it found.

The servers that run online get a row each. The ones that need no sign-in connect together in one click. The rest get a Sign in button that opens the service’s own sign-in. A server you would rather add by hand only needs its address.

If your computer is linked with the runner, you skip the paste. The board lists the servers it found in your apps there, by name and web address, and adds only the ones you choose.

Then any of their read tools can feed a widget. The form runs the tool and shows what the widget will say before anything is added.

Where to find your settings

Claude Desktop: Settings, then Developer, then Edit Config
Claude Code: ~/.claude.json, or .mcp.json in a project
Cursor: ~/.cursor/mcp.json
VS Code: .vscode/mcp.json
Windsurf, now Devin Desktop: ~/.codeium/windsurf/mcp_config.json
Codex: ~/.codex/config.toml
Gemini CLI: ~/.gemini/settings.json

What the board runs on your servers

The board only runs a server’s tools that read. Anything that creates, changes or deletes stays with your AI app. So a widget fed by your Linear server can show your open issues, and it can never close one.

Your keys stay put too. What you paste is read on the page, and only each server’s web address is checked. When you sign in to a server, the board seals that sign-in on its own server and never sends it back.

Servers that are programs on your computer, the ones started with a command, are named and explained instead. A web page cannot start them. Your AI app keeps using them, and it can still send what it finds to the board.

Frequently asked questions

Something else on your mind? Get in touch.

What is an MCP dashboard?

A dashboard whose widgets read from MCP servers, the same servers your AI agent already uses. Context Board connects them once and shows what their read tools return, live while the board is open, beside everything else you track.

Which settings files can it read?

Claude Desktop (Settings, then Developer, then Edit Config), Claude Code (~/.claude.json, or .mcp.json in a project), Cursor (~/.cursor/mcp.json), VS Code (.vscode/mcp.json), Windsurf, now Devin Desktop (~/.codeium/windsurf/mcp_config.json), Codex (~/.codex/config.toml) and Gemini CLI (~/.gemini/settings.json). A plain list of server addresses works too.

What about MCP servers that run on my computer?

A web page cannot start a program on your computer, so the board names those servers and explains why it left them. Your AI app keeps using them, and it can still send what it finds to the board.

Does the board change anything through my servers?

No. The board only runs a remote server's tools that read. Creating, changing and deleting stay with your AI app, so a widget fed by your Linear server can show an issue and never close one.

Do my keys get sent to the board?

No. What you paste is read on the page. The board checks each server's web address, and keys, tokens and anything else in the file never leave the page. An address that may hold its own secret is only checked if you ask.

What does it cost?

Context Board is free for 3 days, then $6 a month. One plan with every connection, widget and theme, your own MCP servers included.

Go deeper

Keep reading

All integrations