MCP Client

Rostyman can connect to external MCP servers, giving you access to tools provided by other services. For example, you can connect to a GitHub server to manage repositories, a Postgres server to query databases, or a filesystem server to read and write files — all from within Rostyman.

What This Does

When you connect to an external MCP server, its tools become available inside Rostyman. You can see what tools each server provides, call them directly, and view the results. Connected server tools are also available to the AI Assistant, which can use them to help with your workflow.

Adding a Server

1. Open the MCP panel (server icon in the right sidebar)
2. Click the Servers tab
3. Click + Add Server
4. Fill in the server details (see below)
5. Click Add & Connect

Rostyman automatically connects to the server after adding it and lists the available tools.

Two Connection Modes

Mode	When to use	Example
Command (stdio)	The server runs as a local process on your machine. Rostyman launches it and communicates via stdin/stdout.	GitHub MCP server, Postgres, Filesystem
URL (HTTP/SSE)	The server is already running somewhere and you connect via a URL.	Remote APIs, cloud-hosted MCP servers

Command Mode (Local Servers)

Most popular MCP servers run locally via the Command mode. You provide the command to launch the server, and Rostyman handles the rest.

Fields

Field	Description	Example
Name	A label for this server (your choice)	GitHub
Command	The executable to run	npx
Arguments	Command-line arguments (space-separated)	-y @modelcontextprotocol/server-github
Environment Variables	Key-value pairs passed to the server process	GITHUB_TOKEN = ghp_xxx...

Step-by-Step: GitHub MCP Server

1. Click + Add Server
2. Set Name to GitHub
3. Set Transport to Command
4. Set Command to npx
5. Set Arguments to -y @modelcontextprotocol/server-github
6. Click + Add Environment Variable
7. Set the key to GITHUB_TOKEN and the value to your GitHub personal access token
8. Click Add & Connect

Once connected, you will see tools like create_issue, search_repositories, get_file_contents, and more.

Step-by-Step: Postgres MCP Server

1. Click + Add Server
2. Set Name to Postgres
3. Set Transport to Command
4. Set Command to npx
5. Set Arguments to -y @modelcontextprotocol/server-postgres postgresql://user:pass@localhost:5432/mydb
6. Click Add & Connect

Once connected, you get a query tool that lets you run SQL queries against your database.

Step-by-Step: Filesystem MCP Server

1. Click + Add Server
2. Set Name to Filesystem
3. Set Transport to Command
4. Set Command to npx
5. Set Arguments to -y @modelcontextprotocol/server-filesystem /path/to/allowed/directory
6. Click Add & Connect

This gives you tools to read, write, search, and list files within the specified directory.

Step-by-Step: Slack MCP Server

1. Click + Add Server
2. Set Name to Slack
3. Set Transport to Command
4. Set Command to npx
5. Set Arguments to -y @modelcontextprotocol/server-slack
6. Add an environment variable: SLACK_BOT_TOKEN = xoxb-xxx...
7. Add an environment variable: SLACK_TEAM_ID = T0xxxxx
8. Click Add & Connect

URL Mode (Remote Servers)

For servers that are already running and accessible via HTTP, use URL mode.

Fields

Field	Description	Example
Name	A label for this server	My Remote API
URL	The server's MCP endpoint	https://mcp.example.com/mcp
Authentication	None or Bearer token	Bearer sk-xxx...

Adding a Remote Server

1. Click + Add Server
2. Set Transport to URL (HTTP)
3. Enter the server URL (the MCP endpoint, typically /mcp)
4. If the server requires authentication, select Bearer Token and enter the token
5. Click Add & Connect

Environment Variables and Security

When you add environment variables (for API keys, tokens, etc.), keep in mind:

• Environment variables are stored locally in your Rostyman database
• Values are masked by default in the Add Server dialog (click the eye icon to reveal)
• Variables are only passed to the specific server process they are configured for
• No environment variables are ever sent to Rostyman's servers or any third party

Tip: For sensitive tokens, consider using short-lived or scoped tokens. For example, create a GitHub token with only the permissions the MCP server needs.

Managing Servers

Connection Status

Each server in the Servers tab shows its current status:

Indicator	Meaning
Green checkmark	Connected and ready
Red X	Connection failed (hover to see the error)
Grey square	Not connected

Connect / Disconnect

• Click Connect to establish a connection to a stopped server
• Click Disconnect to close the connection (the server configuration is kept)
• Enabled servers auto-connect when Rostyman starts

Remove a Server

Click the trash icon next to any server to remove it. This disconnects the server and deletes its configuration.

Viewing Available Tools

When a server is connected, the number of available tools is shown next to the server name. To see the full list of tools, check the Overview tab in the MCP panel, which shows all connected servers and their tool counts.

Connection Troubleshooting

"Command not found" error

The command (e.g., npx, node, python) must be available in your system PATH. If you installed Node.js or Python but the command is not found:

• Windows: Make sure the installer added the tool to your PATH, or provide the full path to the executable
• macOS/Linux: Check that your shell profile exports the correct PATH

"Connection refused" (URL mode)

1. Verify the URL is correct and the server is running
2. Check that the port is not blocked by a firewall
3. If the server requires authentication, make sure you selected Bearer Token and provided the correct token

Server connects but shows 0 tools

1. The server may still be initializing — wait a few seconds and check again
2. Disconnect and reconnect
3. Check the Logs tab for error messages from the server

Server keeps disconnecting

1. Check the Logs tab for error details
2. For Command mode servers, make sure the command does not exit prematurely
3. For URL mode servers, check that the remote server is stable and reachable

Logs

All MCP activity is logged and viewable in the Logs tab of the MCP panel.

What the Logs Show

• INFO: Connection events, server starts/stops
• WARN: Non-critical issues like reconnection attempts
• ERROR: Connection failures, tool call errors
• TOOL: Individual tool calls with direction arrows (request/response)

Searching Logs

Type in the search box at the top of the Logs tab to filter entries. The search matches against the log message text.

Verbose Mode

Enable the Verbose checkbox to see full request and response payloads in tool call logs. This is helpful when debugging but produces more output.

Log Files

Logs are saved to disk as daily files. Click the folder icon in the Logs toolbar to open the log directory in your file manager. Log files are named mcp-YYYY-MM-DD.log and are automatically cleaned up after 30 days.

Clearing Logs

Click Clear in the Logs toolbar to clear the in-memory log view. This does not delete the log files on disk.