metatron

Applies common patterns for the metatron distributed computing language and virtual machine.

3|1|Updated Aug 22, 2025
One-click install
npx skills add https://github.com/phaseshift-studio/metatron --skill metatron-phaseshift-studio
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: metatron
Source: https://github.com/phaseshift-studio/metatron/tree/main/docs/website/skills/metatron
Command: npx skills add https://github.com/phaseshift-studio/metatron --skill metatron-phaseshift-studio

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires websockets, and includes scripts (resource) and references (resource) components.

What problem does it solve? Working effectively with the metatron distributed computing language and VM requires knowing recurring patterns for agent tool execution, MCP server integration, type systems, rewrites, and relational spaces, which are otherwise scattered across the codebase. ## Core Features & Use Cases - Agent Tool Execution: Configure agents with the tool_feature of the /m/llm instset so they can evaluate mtron expressions and call MCP server tools. - MCP Client and Server Patterns: Connect to MCP servers over WebSocket, HTTP, or STDIO transports by converting mcpServer JSON snippets into mcp_client::T instances. - Java Internals References: Consult deep references on the type system, rewrite system, UI architecture, and tbleSpace relational-database integration. - Use Case: An agent needs to execute mtron code against a running VM; this skill provides the agent definition snippet, the mtron_ws_client.py websocket script, and the MCP reference to wire everything together. ## Quick Start Use the metatron skill to show me how to connect an agent to an MCP server and execute mtron expressions over the websocket endpoint.

Frequently Asked Questions about metatron

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I let an agent execute mtron expressions?

Attach the tool_feature from the /m/llm instset to the agent definition, for example tool_feature::[tool=>[!*eval]]. This gives the spawned agent access to mtron instructions and MCP server tools through a general-purpose eval instruction.

How do I connect to an MCP server from metatron?

Convert the server's published mcpServer JSON snippet into an mcp_client::T via the path str::T => json::T => mcp_client::T. WebSocket, HTTP stream, and STDIO transports are supported, and the client's tool field populates with callable tool entries.

What transports does the metatron MCP client support?

The mcp_client::T supports websocket, streamable HTTP, and STDIO transports. For STDIO servers you supply a command array, while HTTP and WebSocket servers are configured with a host URL.

How do I evaluate mtron code over a websocket?

Use the scripts/mtron_ws_client.py Python script, which connects to a metatron websocket endpoint such as ws://127.0.0.1:8999 and submits expressions through the /m/web/inst/doc instruction. It requires the websockets Python package.

What Java internals does this skill document?

The references cover the type system (Type interface, MType factory, coefficients, LCD), the rewrite system (Rewriter, RewriteBuilder, Code.rewrite loop), the UI widget architecture, and tbleSpace relational-database integration with SQL pushdown.