Can I use Claude chat connectors in Claude Code?

Not directly. Claude chat connectors are set up through your account settings and are available in the browser interface. Claude Code uses a separate local MCP configuration. Some tools offer both a Claude connector and an MCP server, so you can connect to the same service from both clients.

Do I need an API key to add an MCP server to Claude Code?

Usually yes. Most MCP servers authenticate with an API key or token for the connected service. Store the key as an environment variable rather than hardcoding it in the configuration file.

Does Claude Code support OAuth connections?

Claude Code supports MCP servers that handle their own authentication, which can include OAuth flows. The setup is still done through the local configuration file rather than through a browser settings panel.

What is the difference between Claude Code and Claude in VS Code?

Claude Code is a command-line tool that runs in your terminal. Claude extensions for VS Code add AI assistance directly to the editor interface. Both can use MCP connections, but the configuration paths differ. Check the documentation for whichever tool you are using.

Claude Connectors in Claude Code: Setup and Access Checks

Claude Code connects to external tools differently from Claude in the browser. Here is how connector setup works in Claude Code, how it compares to the chat interface, and how to give both clients one shared context record.

October 23, 2026Updated October 21, 20266 min read

Claude Code connects to external tools through MCP (Model Context Protocol) — a standard way for an AI to use an outside tool — configured locally in a JSON file on your machine. This is different from the connectors in Claude's browser-based chat interface, which are set up through account settings using OAuth. Understanding which path applies to your workflow prevents wasted setup time. Tools like fenbs provide one shared context record that both Claude Code and Claude chat can read, so the same project context is available regardless of which client you are using.

Claude Code is a coding assistant that runs in your terminal and integrates with your local development environment. It reads your codebase, runs commands, and can connect to external tools like issue trackers, documentation sources, and project management systems through MCP servers.

How Claude Code connections work

Claude Code uses a local MCP configuration file — typically .claude/settings.json or a project-level CLAUDE.md paired with a server configuration — to know which external tools it can reach. Each tool is an MCP server: a small process that Claude Code calls to read or write data in an external system.

Unlike Claude's browser interface, which handles OAuth in a pop-up window, Claude Code's MCP connections are configured with credentials or tokens set as environment variables or in the configuration file. This means the setup is done by the developer before starting a session, not by the end user clicking Authorize in a browser.

Setting up an MCP server for Claude Code

To add a connector in Claude Code, add the MCP server to your Claude Code settings. For project-level connections, this belongs in a settings file in your project root. For personal connections available across all projects, add them to your user-level Claude Code settings.

Identify the MCP server for the tool you want to connect. Many tools publish an official MCP server package.
Install the server package, following the tool's installation instructions.
Add the server configuration to your Claude Code settings file with the correct command and environment variable references for the API key or token.
Store the API key or token as an environment variable, not as a literal string in the config file.
Test by asking Claude Code to list the tools available from the new server.

For the current Claude Code MCP configuration format, see the Claude Code documentation and Anthropic's official guides for Claude Code setup.

How this differs from Claude chat connectors

Claude in the browser has a connector settings panel where you authorize connections using OAuth. The user clicks through a consent flow, approves the scopes, and the connector becomes available in that account. No local configuration is needed.

Claude Code connectors require local configuration and typically use API keys rather than OAuth flows. This is appropriate for a developer tool where the environment is already controlled. It is less appropriate for non-technical users who expect a one-click setup.

Claude chat connectors versus Claude Code MCP connections
	Claude chat	Claude Code
Setup	Account settings, OAuth browser flow	Local config file, API key or token
Auth method	OAuth with user consent	Environment variable or token in config
Who configures it	End user, in the browser	Developer, before the session starts
Available to	Anyone with the account	That machine and user profile
Best for	Browser-based workflows, non-technical users	Code editing, terminal workflows, local dev environments

Access checks for Claude Code connections

Because Claude Code connections use API keys stored locally, review the permission scope of each key before configuring it. A key with write access to a production system in a Claude Code configuration is as powerful as that same key used manually. Apply the same access controls you would use for any service account.

Keep credentials in environment variables or a secrets manager, not in files that might be committed to source control. Add configuration files that contain credentials to .gitignore. Rotate keys if they are exposed.

One shared context for Claude Code and Claude chat

When you use both Claude Code for development work and Claude chat for design or research tasks, each session can end up with a different picture of the project. Decisions made in one client are not automatically visible to the other.

Connecting both clients to the same shared context record means each session starts from the same facts: the current rules, what is in progress, what is waiting on a person. The record lives outside both clients, so switching between them does not split the work.

How fenbs helps

One context record for both Claude Code and Claude chat

fenbs connects to Claude Code as an MCP server and to Claude chat as a connector. Both read the same project context: rules, tasks, decisions, where things stand. A developer finishing a feature in Claude Code and a designer reviewing copy in Claude chat both work from the same record. Connect at Integrations and follow the guide for each client.

Create a free account

Free for one person · No credit card · Works with Claude, ChatGPT, Cursor and more

How to do this with fenbs

Create a free account and add your development project.
Connect Claude Code by following the Claude Code setup guide at Integrations — it configures fenbs as an MCP server.
Connect Claude chat through your Claude account settings using the OAuth flow at the same page.
Ask both clients to read the project context before starting work: rules, tasks, where things stand.
Before ending any session, ask the AI to save progress so the next session — in either client — starts from the current state.

For Claude Code installation and setup basics, see How to Install Claude Code and the Claude Code docs overview. For Claude connector types in general, see Claude Connectors Explained. More guides are in the connect-ai topic.