JCode

v1.0.0Coding Agentsstable

Coding Agent Harness

aiclaudeclicoding-agentllm
Share:
6,444
Stars
0
Downloads
0
Weekly
0/5

What is JCode?

JCode is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to coding agent harness

Coding Agent Harness

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Coding Agent Harness

Use Cases

Coding agent harness
LLM integration
Code generation
1jehuang

Maintainer

LicenseMIT
Languagerust
Versionv1.0.0
UpdatedMay 22, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx jcode

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use JCode

jcode is a next-generation AI coding agent harness built in Rust that prioritizes extreme performance, low memory footprint, and multi-session scalability. It features a semantic memory system that embeds every conversation turn as a vector and automatically retrieves relevant context, a side panel for real-time file and diff viewing, inline Mermaid diagram rendering, and built-in login flows for Claude, OpenAI, Gemini, GitHub Copilot, Azure OpenAI, and Fireworks. Developers running parallel agentic coding sessions use jcode because it uses as little as 27 MB of RAM per session compared to hundreds of megabytes for other agent CLIs.

Prerequisites

  • macOS or Linux (Windows supported via manual install — see GitHub releases)
  • An API key or subscription for at least one supported LLM provider (Claude, OpenAI, Gemini, GitHub Copilot, Azure OpenAI, or Fireworks)
  • curl for the one-line installer, or access to GitHub releases for manual binary download
  • Optional: the Handterm terminal for smooth partial-line scrolling support
1

Install jcode on macOS or Linux

Run the one-line installer script to download and install the latest jcode binary for your platform.

curl -fsSL https://raw.githubusercontent.com/1jehuang/jcode/master/scripts/install.sh | bash
2

Log in to your preferred LLM provider

jcode supports OAuth-based login for multiple providers. Run the login command for the provider you want to use — no manual API key copy-pasting required for supported providers.

# Choose one:
jcode login --provider claude
jcode login --provider openai
jcode login --provider gemini
jcode login --provider copilot
jcode login --provider azure
jcode login --provider fireworks
3

Start a coding session

Launch jcode in your project directory to begin an interactive agent session. The agent will have access to your files and shell.

jcode
4

Use the side panel and alignment options

Ask the agent to load a file into the side panel for real-time viewing. Toggle centered layout with Alt+C or the /alignment command.

5

Configure the MCP server entry for Claude Desktop

To use jcode as an MCP server from within Claude Desktop, add it to your claude_desktop_config.json.

6

Run multiple sessions in parallel

Open additional terminals and run jcode in each to take advantage of its low per-session memory usage (~10 MB incremental RAM per session with local embeddings enabled).

JCode Examples

Client configuration

Claude Desktop config to launch jcode as an MCP server.

{
  "mcpServers": {
    "jcode": {
      "command": "npx",
      "args": ["jcode"]
    }
  }
}

Prompts to try

Example tasks to try in a jcode coding session.

- "Refactor the authentication module to use async/await throughout"
- "Load src/main.rs into the side panel and explain what each function does"
- "Search my memory for previous discussions about the database schema"
- "Run the test suite and fix any failing tests"
- "Draw a Mermaid sequence diagram of the request lifecycle in this app"
- "What did we decide about error handling in our last session?"

Troubleshooting JCode

Install script fails with 'permission denied'

The installer writes to /usr/local/bin by default. Run with sudo if needed: `curl -fsSL https://raw.githubusercontent.com/1jehuang/jcode/master/scripts/install.sh | sudo bash`, or set a custom install prefix via the INSTALL_DIR environment variable.

jcode login --provider claude opens a browser but authentication never completes

Ensure the local callback server (typically localhost:7777) is not blocked by a firewall. Allow it in your OS firewall settings and retry the login command.

Memory retrieval returns irrelevant results

The semantic memory system requires sufficient conversation history to build meaningful embeddings. Memory consolidation runs automatically every K turns and at session end. Use the explicit memory search tool (`/memory search <query>`) for immediate retrieval without waiting for background consolidation.

Frequently Asked Questions about JCode

What is JCode?

JCode is a Model Context Protocol (MCP) server that coding agent harness It connects AI assistants to external tools and data sources through a standardized interface.

How do I install JCode?

Follow the installation instructions on the JCode GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with JCode?

JCode works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is JCode free to use?

Yes, JCode is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "jcode": { "command": "npx", "args": ["-y", "jcode"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use JCode?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides