pctx

Overview

pctx serves as an execution layer for agentic tool calls, automatically converting agent tools and MCP servers into code that runs in secure Deno sandboxes. It provides a unified interface to multiple upstream MCP servers, handling authentication and exposing tools through Code Mode, which reduces token usage by up to 98.7% for multi-step operations by replacing sequential tool calls with code execution in isolated environments.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

Comparable tools

Installation

# Homebrew
brew install portofcontext/tap/pctx

# cURL
curl --proto '=https' --tlsv1.2 -LsSf https://raw.githubusercontent.com/portofcontext/pctx/main/install.sh | sh

# npm
npm i -g @portofcontext/pctx

{
  "mcpServers": {
    "pctx": {
      "command": "pctx",
      "args": ["mcp", "start", "--stdio"]
    }
  }
}

FAQ

What is Code Mode?

Code Mode replaces sequential tool calling with code execution. Instead of an agent calling tools one at a time, it writes code that executes in a sandbox, significantly reducing token usage for multi-step operations.

How does pctx ensure security?

LLM-generated code runs in an isolated Deno sandbox with restricted access to filesystem, environment, and network (beyond allowed hosts). MCP clients are authenticated in pctx so LLMs never see authentication credentials.

On Hacker News

Recent discussion from the developer community.

• Show HN: Pctx – OSS to build efficient AI agents with Code Mode▲ 1
  Story by pmkelly4444 · 2025-11-19
• Show HN: Python Local Sandbox Code Execution (Podman and Uv)▲ 4
  Story by pmkelly4444 · 2025-12-22

Compare pctx with