MCP Explorer

App Settings

Mon, 01 Jan 0001 00:00:00 +0000

Overview

The Settings page provides global configuration for MCP Explorer — themes, sensitive data protection rules, and data import/export.

The Settings page groups configuration into cards: Theme, Security, and Data Management.

Themes

MCP Explorer ships with 10 built-in themes. Select any theme — your choice is persisted automatically.

Theme	Style
Command Dark	Dark, high-contrast terminal feel
Command Light	Clean light counterpart
Nord	Arctic blue palette
Dracula	Classic dark purple
Catppuccin Mocha	Warm dark with pastel accents
Solarized Light	Warm off-white, reduced glare
GitHub Dark	Familiar dark mode
GitHub Light	Classic GitHub light
Material Dark	Google Material dark
Material Light	Google Material light

Sensitive Data Protection

MCP Explorer detects and masks sensitive values using a combination of heuristics and configurable regex patterns.

Browsing & Invoking Tools

Mon, 01 Jan 0001 00:00:00 +0000

Overview

The Tools page lists every tool exposed by your active MCP connections. Tools are the primary way MCP servers expose functionality — from web search to file I/O, database queries to API calls.

Select a connection on the left to load its tools. Each tool shows its name and description. The active tool’s parameter form opens in the right panel.

Building Workflows

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Workflows let you chain MCP tool calls into repeatable, parameterised sequences. Define the steps once, run them on demand, track history, and even load-test your server with concurrent executions.

The Workflows list. Click New Workflow to create your first workflow, or Import to load one from a JSON file.

Creating a Workflow — Step-by-Step Example

The walkthrough below creates an Echo and Lookup workflow that:

Chat Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

The Chat view gives you a streaming conversation interface backed by any LLM you’ve configured. What makes it powerful is automatic MCP tool calling — the LLM can invoke any tool from your active connections mid-conversation, and MCP Explorer handles the round-trip transparently.

Starting a Chat

Click New Session in the left sidebar to create a fresh conversation
Select which Connections to make available for tool calling from the multiselect dropdown
Choose your Model (defaults to GPT-4o)
Type your message and press Ctrl+Enter or click the Send button

A new chat session with the mcp server connection selected. The connection badge appears next to the model selector, confirming the LLM has access to all MCP tools from that server.

Configuring Models

Mon, 01 Jan 0001 00:00:00 +0000

Overview

The Models page is where you configure the LLMs that power the Chat view. MCP Explorer supports any OpenAI-compatible API endpoint.

The AI Models page lists all configured LLMs. Each entry shows its friendly name, provider badge (e.g. AzureAIFoundry), model ID, endpoint URL, and masked API key. The default model is highlighted — it is pre-selected in the Send to LLM picker.

Keyboard Shortcuts

Mon, 01 Jan 0001 00:00:00 +0000

Command Palette

Press ⌘K (macOS) or Ctrl+K (Windows/Linux) from anywhere in the app to open the command palette.

The command palette provides instant keyboard-first navigation to any page or action. Type to fuzzy-search.

Type to fuzzy-search and press Enter to navigate. Press Esc to close.

Global Shortcuts

Shortcut	Action
`⌘K` / `Ctrl+K`	Open command palette
`Esc`	Close dialogs / command palette

Chat Shortcuts

Shortcut	Action
`Enter`	Send message
`Shift+Enter`	Insert newline
`/` (at start of input)	Open slash commands
`Ctrl+Enter`	Send (alternate)

Slash Commands (Chat)

Type / in the chat input to access:

Managing Connections

Mon, 01 Jan 0001 00:00:00 +0000

Overview

The Connections page is your hub for all MCP server connections. From here you can add new servers, edit existing ones, group them logically, and connect or disconnect at any time.

The Connections page lists all saved MCP servers with live health indicators, transport type, and connection toggles. Use the toolbar to add, import, or export connections.

Quick Start

Mon, 01 Jan 0001 00:00:00 +0000

Prerequisites

Docker 20.10+ installed and running
A terminal / command prompt

Option 1 — Single Container (`docker run`)

The fastest way to get started. One command, one container.

Using Elicitations

Mon, 01 Jan 0001 00:00:00 +0000

What are Elicitations?

Some MCP tools need additional user input mid-execution — for example, asking which of several options to proceed with, or requesting a confirmation before a destructive operation. This is the MCP elicitation feature.

MCP Explorer surfaces these requests as inline dialogs that pause the tool execution and wait for your response.

How Elicitations Work

sequenceDiagram
 participant U as User
 participant App as MCP Explorer
 participant T as MCP Tool

 U->>App: Invoke tool
 App->>T: Tool call
 T-->>App: Elicitation request (needs input)
 App-->>U: Show inline input dialog
 U->>App: Provide response
 App->>T: Continue with user input
 T-->>App: Tool result
 App-->>U: Display result

Responding to an Elicitation

When a tool makes an elicitation request, an inline prompt appears within the tool card:

Using Prompts

Mon, 01 Jan 0001 00:00:00 +0000

Overview

MCP Prompts are parameterised message templates that MCP servers expose. They let you capture common workflows — code review patterns, data analysis templates, document generators — and invoke them with a few inputs.

Select a connection to load its prompts. Each prompt shows its name, description, and argument count.

Executing a Prompt

Click a prompt to open its detail panel
Fill in any required arguments

Each prompt argument has a label, description, and validation indicator. Fill in the values and click Execute.

Using Resource Templates

Mon, 01 Jan 0001 00:00:00 +0000

What are Resource Templates?

Resource Templates are URI patterns that MCP servers expose with placeholders — for example file:///{path}, db:///{table}/{id}, or api:///{endpoint}?query={q}. By filling in the placeholders, you resolve the template to a concrete URI and fetch the corresponding resource.

They’re distinct from static Resources in that they’re dynamic — the same template can produce unlimited concrete resources.

Finding Templates

The Resource Templates tab is accessible from the Resources page (tab switcher at the top). Templates are listed with:

Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Architecture Overview

MCP Explorer follows a clean architecture pattern with a Vue 3 SPA frontend and an ASP.NET Core backend.

flowchart TB
 subgraph BROWSER["🌐 Browser"]
 SPA["⚡ Vue 3 · Vite · PrimeVue 4\nSingle-Page Application"]
 end

 subgraph SINGLE["🐳 Single Container (docker run)"]
 GW["🔀 YARP Gateway :8080\nStatic SPA host + /api/** proxy"]
 API["⚙️ ASP.NET Core API :5000"]
 GW -->|reverse proxy| API
 end

 subgraph DATA["💾 Persistent Data"]
 FILE["/app/data/data.json\nConnections · Models · Workflows\nSettings · Chat History"]
 end

 subgraph MCP["🔌 MCP Servers"]
 S1["Server A (stdio)"]
 S2["Server B (SSE)"]
 S3["Server C (HTTP)"]
 end

 BROWSER -->|HTTP/SSE| GW
 API -->|read/write| FILE
 API -->|MCP protocol| S1
 API -->|MCP protocol| S2
 API -->|MCP protocol| S3

Deployment Modes

Single Container

The default deployment — everything in one image.

Configuration

Mon, 01 Jan 0001 00:00:00 +0000

Environment Variables

MCP Explorer is configured via environment variables. When using Docker Compose, place these in a .env file alongside docker-compose.yml.

Variable	Default	Description
`MCP_CLIENT_NAME`	`mcp-explorer`	Name sent in the `User-Agent` header and MCP `ClientInfo` when connecting to MCP servers
`ASPNETCORE_ENVIRONMENT`	`Production`	Set to `Development` for verbose logging

Example `.env`

 # .env
MCP_CLIENT_NAME=acme-mcp-explorer

Persistent Data

MCP Explorer stores all state — connections, models, workflows, chat history, and settings — in a single JSON file at /app/data/data.json inside the container.

Data Guard

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Data Guard is MCP Explorer’s built-in sensitive data protection layer. It intercepts all MCP communications — tool inputs, tool outputs, prompt content, and chat messages — and automatically detects, masks, and protects sensitive values before they appear in the UI or are stored.

Detection uses three complementary engines that run in combination:

Engine	Always on?	Description
Regex patterns	✅ Yes	Detects common secret formats: API keys, bearer tokens, passwords, connection strings
Heuristic scanning	✅ Yes	Flags high-entropy alphanumeric tokens that look like secrets
AI detection	⚙️ Optional	Uses an LLM to analyse values in context — catches secrets that evade pattern matching

Detected values are masked as ●●●●●●●● in the UI with a reveal toggle, and are encrypted at rest.

Environment Variables

Mon, 01 Jan 0001 00:00:00 +0000

Overview

MCP Explorer is configured entirely via environment variables — there are no required config files to edit. Copy .env.example to .env, fill in your values, and pass it to Docker or Docker Compose.

 # Docker Compose (reads .env automatically)
cp .env.example .env
docker compose up -d

# docker run (pass vars individually)
docker run -p 8090:8080 \
 -e MCP_CLIENT_NAME=my-app \
 -e MCP_DATA_PATH=/data \
 -v /path/to/data:/data \
 ghcr.io/your-org/mcp-explorer:latest

info: In Docker Compose, variables in .env are interpolated into docker-compose.yml automatically. Variables prefixed VITE_ are build-time only — they are baked into the static JS bundle and cannot be changed at runtime.

MCP Explorer

App Settings

Overview link

Themes link

Sensitive Data Protection link

Browsing & Invoking Tools

Overview link

Building Workflows

Overview link

Creating a Workflow — Step-by-Step Example link

Chat Overview

Overview link

Starting a Chat link

Configuring Models

Overview link

Keyboard Shortcuts

Command Palette link

Global Shortcuts link

Chat Shortcuts link

Slash Commands (Chat) link

Managing Connections

Overview link

Quick Start

Prerequisites link

Option 1 — Single Container (docker run) link

Using Elicitations

What are Elicitations? link

How Elicitations Work link

Responding to an Elicitation link

Using Prompts

Overview link

Executing a Prompt link

Using Resource Templates

What are Resource Templates? link

Finding Templates link

Architecture

Architecture Overview link

Deployment Modes link

Single Container link

Configuration

Environment Variables link

Example .env link

Persistent Data link

Data Guard

Overview link

Environment Variables

Overview link

Overview

Themes

Sensitive Data Protection

Overview

Overview

Creating a Workflow — Step-by-Step Example

Overview

Starting a Chat

Overview

Command Palette

Global Shortcuts

Chat Shortcuts

Slash Commands (Chat)

Overview

Prerequisites

Option 1 — Single Container (`docker run`)

What are Elicitations?

How Elicitations Work

Responding to an Elicitation

Overview

Executing a Prompt

What are Resource Templates?

Finding Templates

Architecture Overview

Deployment Modes

Single Container

Environment Variables

Example `.env`

Persistent Data

Overview

Overview