Get the FREE Ultimate OpenClaw Setup Guide →

1mcp

Let your agent write code and execute code directly in the browser with WASM

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio buremba-1mcp http https://mcp.sentry.dev/mcp

How to use

1mcp lets agents compose MCP tool calls and run code safely via WASM, enabling chained tool execution with strong sandboxing to reduce token usage. The server exposes a curated set of tools by default (e.g., filesystem access, and integration points via HTTP transports to external MCPs). Developers can write JavaScript/TypeScript code that chains MCP tool calls, and 1mcp handles dispatch, sandboxing, and retries automatically. You can integrate with the @onemcp/ai-sdk to convert AI SDK tools into sandboxed MCP tools for a unified execution surface. To use, install the server globally, initialize configuration, and start serving; then connect your AI agent to the exposed MCP tools and execute tool chains in a single capsule when possible to minimize token cost.

How to install

Prerequisites:

  • Node.js and npm installed on your system
  • Access to the npm registry (Internet)

Steps:

  1. Install the 1mcp CLI globally npm install -g 1mcp

  2. Initialize configuration (creates 1mcp.config.json or uses default path) 1mcp init

  3. Start the MCP server 1mcp serve

  4. (Optional) Customize 1mcp.config.json to tune mcps, policy, and environment

    • Refer to the Configuration Format in the README for details about mcps, policy, and session settings
  5. Integrate with your AI workflow or SDKs once the server is running

    • Use the exposed MCP tools via the AI SDK bridge or direct API as described in the README

Additional notes

Tips and considerations:

  • The available mcps include a filesystem tool (exposed via npx) and external MCPs accessed over HTTP. For HTTP-based MCPs, the transport is defined as http in the mcps config.
  • Security policies (network, filesystem, and runtime limits) are configurable per session. Ensure your policy aligns with the tools you enable.
  • CLI flags override environment variables, which override config defaults. Common settings include timeout (TIMEOUT_MS), max memory (MAX_MEMORY_MB), and max stdout bytes (MAX_STDOUT_BYTES).
  • The 1mcp CLI offers init and serve commands to bootstrap and run the server. You can adjust the 1mcp.config.json to tailor dependencies, allowed domains, and resource limits.
  • If you encounter issues with the filesystem tool, verify that the sandbox path (e.g., /private/tmp) is writable within the configured policy and that the host environment has appropriate permissions.

Related MCP Servers

Sponsor this space

Reach thousands of developers