MCPico
MCP proxy that bundles flat tool lists into hierarchical subcommand groups to reduce context token usage, supporting multi-server aggregation and auto-generated help from tool schemas.
README
MCPico
MCP proxy that bundles flat tool lists into hierarchical subcommand groups.
MCPico (MCP + "ico" = tiny) wraps upstream MCP servers, grouping their tools into discoverable subcommand-based tools. One tool per group, not one per tool. The help subcommand auto-generates rich documentation from upstream schemas.
The Problem
MCP servers expose tools as a flat list. Every tool costs context tokens. A filesystem server exposes 14+ separate tools — the model sees all of them, all the time, even when it only needs one.
Some clients add "tool search" as a workaround. But searching requires the model to proactively look for tools it doesn't know exist. No structural signal about which tools relate to each other.
MCPico's Solution
Group related tools under a single entry point. The model sees 9 groups instead of 14 tools. Discovery is built-in via help:
→ filesystem tools:
14 tools → 9 groups: read, write, edit, create, list, directory, move, search, get
Features
- Tool bundling — Groups tools by prefix (configurable separator), collapsing flat tool lists
- Auto-generated help — Each group's
helpsubcommand is built from upstream tool schemas - Multi-server aggregation — Proxy multiple upstream MCP servers through one interface
- Dual transport — Supports both stdio and Streamable HTTP (SSE) upstream servers
- Configurable timeouts — Per-server connection timeout with sensible default (30s)
- Resource & prompt passthrough — Namespaced to avoid collisions across servers
Usage
Install
npm install -g mcplico
Configure
Create mcplico.json:
{
"servers": [
{
"name": "filesystem",
"transport": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"]
}
}
]
}
Run
mcplico
Connect your MCP client
Add MCPico as a server in your MCP client config:
{
"mcpServers": {
"mcplico": {
"command": "mcplico",
"args": ["--config", "/path/to/mcplico.json"]
}
}
}
How it works
- Connect to upstream MCP servers
- Discover their tools (
tools/list) - Group tools by prefix (configurable separator, default
_)filesystem_read_file,filesystem_write_file→ groupfilesystem
- Register each group as a single MCP tool with a
commandstring argument - Forward tool calls by parsing
<subcommand> {"key":"value"}and proxying to upstream - Generate help dynamically from original tool schemas
Command format
<subcommand> {"arg1":"val1","arg2":"val2"}
Examples:
help— see all subcommands and their parametersread_file {"path":"/tmp/hello.txt"}— call a specific toolwrite_file {"path":"/tmp/out.txt","content":"hello"}— with arguments
Multi-server aggregation
MCPico can proxy multiple upstream servers simultaneously:
{
"servers": [
{
"name": "filesystem",
"transport": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]
}
},
{
"name": "github",
"transport": {
"type": "sse",
"url": "https://mcp-github.example.com/mcp"
}
}
]
}
Groups from different servers are merged if they share a prefix. Otherwise each server's tools appear as separate groups.
Configuration
| Field | Type | Default | Description |
|---|---|---|---|
servers |
ServerConfig[] |
required | Upstream MCP servers to proxy |
separator |
string |
"_" |
Separator for prefix-based tool grouping |
groups |
object |
{} |
Explicit group overrides ({ "group": ["tool1","tool2"] }) |
ServerConfig
| Field | Type | Required | Description |
|---|---|---|---|
name |
string |
yes | Friendly name / group namespace |
transport |
TransportConfig |
yes | How to connect to the upstream server |
connectTimeoutMs |
number |
no | Connection timeout in ms (default: 30000) |
TransportConfig (stdio)
| Field | Type | Required | Description |
|---|---|---|---|
type |
"stdio" |
yes | Transport type |
command |
string |
yes | Executable to spawn |
args |
string[] |
no | Command-line arguments |
env |
object |
no | Environment variables |
cwd |
string |
no | Working directory |
TransportConfig (SSE / Streamable HTTP)
| Field | Type | Required | Description |
|---|---|---|---|
type |
"sse" |
yes | Transport type |
url |
string |
yes | Full URL to MCP Streamable HTTP endpoint |
Development
npm install
npm run build # TypeScript compilation
npm test # Run tests (77 tests, vitest)
npm run dev # Run directly with tsx
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.