Skip to main content
Glama
Datron

Jilebi

by Datron
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Rust
Local

A powerful, plugin-based MCP (Model Context Protocol) server that extends AI assistants with custom tools, resources, and prompts.

Table of Contents

Related MCP server: jlbloomer-mcp-server

Overview

Jilebi is an MCP server implementation that enables AI assistants to interact with external systems through a modular plugin architecture. It provides a standardized way to expose tools, resources, and prompts to language models while maintaining security through a granular permission system.

Key Features:

  • Cross-platform support (Windows, Linux, macOS)

  • Plugin-based extensibility

  • Secure permission model for file system and network access

  • Persistent state management for plugins

  • Compatible with major AI development environments

Quick Install

The fastest way to install Jilebi is using our installation scripts:

Linux & macOS

Using curl:

curl -fsSL https://mcp.jilebi.ai/install.sh | bash

Using wget:

wget -qO- https://mcp.jilebi.ai/install.sh | bash

Windows (PowerShell)

irm https://mcp.jilebi.ai/install.ps1 | iex

Or with the full command:

Invoke-RestMethod https://mcp.jilebi.ai/install.ps1 | Invoke-Expression

What the installer does

  • Detects your platform and architecture automatically

  • Downloads the appropriate binary

  • Installs to ~/.jilebi/bin (Unix) or %LOCALAPPDATA%\jilebi\bin (Windows)

  • Adds Jilebi to your PATH

  • Optionally installs recommended plugins (memory, sequential-thinking)

Uninstall

To completely remove Jilebi from your system:

Linux & macOS

Using curl:

curl -fsSL https://mcp.jilebi.ai/uninstall.sh | bash

Using wget:

wget -qO- https://mcp.jilebi.ai/uninstall.sh | bash

Windows (PowerShell)

irm https://mcp.jilebi.ai/uninstall.ps1 | iex

Or with options:

# Skip confirmation prompts
irm https://mcp.jilebi.ai/uninstall.ps1 -OutFile uninstall.ps1; .\uninstall.ps1 -Force

# Keep plugins and data
irm https://mcp.jilebi.ai/uninstall.ps1 -OutFile uninstall.ps1; .\uninstall.ps1 -KeepData

What the uninstaller removes

  • Jilebi binary and installation directory

  • PATH entry from shell profile (Unix) or user environment (Windows)

  • Optionally: plugins, data, and configuration files

Manual Download

Platform

Architecture

Download

Windows 10/11

x86_64

Download

Linux

x86_64

Download

macOS

Apple Silicon

Download

macOS

Intel

Download

Manual Installation

Windows

# 1. Unzip the downloaded file to your desired location

# 2. Install recommended plugins
.\jilebi.exe plugins add memory
.\jilebi.exe plugins add sequential-thinking

# 3. Start the MCP server
.\jilebi.exe stdio

Linux

# 1. Unzip the downloaded file to your desired location

# 2. Grant execution access
chmod +x jilebi

# 3. Install recommended plugins
./jilebi plugins add memory
./jilebi plugins add sequential-thinking

# 4. Start the MCP server
./jilebi stdio

macOS

# 1. Unzip the downloaded file to your desired location

# 2. Grant execution access
chmod +x jilebi

# 3. Remove quarantine attribute (required for unverified developers)
xattr -d com.apple.quarantine /path/to/jilebi

# 4. Install xz dependency
brew install xz

# 5. Install recommended plugins
./jilebi plugins add memory
./jilebi plugins add sequential-thinking

# 6. Start the MCP server
./jilebi stdio

Find more plugins at github.com/datron/jilebi-plugins

Host Configuration

Jilebi integrates with various AI development environments. Below are configuration examples for supported hosts.

Claude Desktop

Config Location:

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "jilebi": {
      "command": "jilebi",
      "args": ["stdio"],
      "env": {}
    }
  }
}

Claude Code

Add to .claude.json:

{
  "mcpServers": {
    "jilebi": {
      "command": "jilebi",
      "args": ["stdio"],
      "env": {}
    }
  }
}

Cursor

Edit ~/.cursor/mcp.json:

{
  "mcpServers": {
    "jilebi": {
      "command": "jilebi",
      "args": ["stdio"],
      "env": {}
    }
  }
}

VS Code

Add to your mcp.json:

{
  "servers": {
    "jilebi": {
      "type": "stdio",
      "command": "jilebi",
      "args": ["stdio"]
    }
  }
}

Zed

Edit ~/.config/zed/settings.json:

{
  "context_servers": {
    "jilebi": {
      "command": "jilebi",
      "args": ["stdio"],
      "enabled": true,
      "source": "custom",
      "env": {}
    }
  }
}

Codex CLI

Add to ~/.codex/config.toml:

[mcp_servers.jilebi]
command = "jilebi"
args = ["stdio"]

Opencode

Edit ~/.config/opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "jilebi": {
      "type": "local",
      "command": ["jilebi", "stdio"],
      "enabled": true,
      "environment": {}
    }
  }
}

Plugin System

Plugins extend Jilebi with custom functionality through three main components:

Component

Description

Tools

Functions that models can invoke to interact with external systems

Resources

Read-only data sources for context (files, schemas, documentation)

Prompts

Pre-defined conversation templates with argument interpolation

Managing Plugins

# Install a plugin
jilebi plugins add <plugin-name>

# Create a new plugin
jilebi plugins create

# Setup/reconfigure a plugin
jilebi plugins setup <plugin-name>

# View plugin logs
jilebi plugins log <plugin-id>

Creating Plugins

Quick Start

jilebi plugins create

Follow the prompts to bootstrap a new plugin project.

Plugin Manifest

Plugins are defined using a TOML manifest file. Here is an example structure:

# Metadata
name = "my-plugin"
version = "1.0.0"
homepage = "github.com/user/my-plugin"
creator = "Your Name"
contact = "you@example.com"

# Environment Variables
[env]
API_URL = { schema = { type = "string" }, default = "https://api.example.com" }

# Secrets (no defaults allowed)
[secrets]
API_KEY = { schema = { type = "string" } }

# Tools
[tools.my-tool]
name = "my-tool"
description = "Description of what this tool does"
input_schema = { type = "object", properties = { param = { type = "string" } }, required = ["param"] }
function = "my_tool_function"

[tools.my-tool.permissions]
hosts = ["https://api.example.com"]

# Resources
[resources.my-resource]
name = "my-resource"
description = "Description of this resource"
mime_type = "application/json"
function = "get_resource"

# Prompts
[prompts.my-prompt]
name = "my-prompt"
description = "A helpful prompt template"
arguments = [
  { name = "topic", description = "The topic to discuss", required = true }
]
messages = [
  { role = "user", content = { type = "text", content = "Tell me about {{topic}}" } }
]

Permission Types

Plugins can request the following permissions:

Permission

Description

hosts

Allowed host domains for network requests

urls

Specific allowed URLs

read_files

File paths the plugin can read

write_files

File paths the plugin can write

read_dirs

Directories the plugin can read from

write_dirs

Directories the plugin can write to

Use "user_defined" to prompt the user for a custom value during setup.

Built-in Functions

Plugins have access to:

State Management:

  • setState(env, key, value) - Persist data across sessions

  • getState(env, key) - Retrieve persisted data

  • deleteState(env, key) - Remove persisted data

File System (Deno API):

  • Directory operations: mkdir, readDir

  • File operations: readFile, writeFile, readTextFile, writeTextFile

  • File manipulation: stat, copyFile, rename, remove

  • Temporary files: makeTempDir, makeTempFile

  • And more...

Logging and Debugging

Log Locations

Platform

Path

Windows

C:\Users\<user>\AppData\Roaming\jilebi\jilebi-server\data\logs

Linux

~/.local/share/jilebi-server/logs/

macOS

~/Library/Application Support/ai.jilebi.jilebi-server/logs

Viewing Logs

# View Jilebi server logs
jilebi log

# View plugin-specific logs
jilebi plugins log <plugin-id>

Writing Logs in Plugins

console.log("Info message");
console.warn("Warning message");
console.error("Error message");
console.debug("Debug message");

Links

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers