Skip to main content
Glama
POEM MCP Server

POEM MCP Server

Official
by openpoem

POEM MCP Server

MCP server for POEM (Pseudo-code Oriented Executable Markup).

Write specs once, get code in any language.

Tools

Tool

What it does

poem_read

Parse a .poem file into structured elements

poem_validate

Check syntax and naming conventions

poem_translate

Prepare a spec for translation to any language

Related MCP server: IntentForge MCP Server

Install

Clone and build:

git clone https://github.com/openpoem/poem-mcp.git
cd poem-mcp
npm install && npm run build

Configure

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "poem": {
      "command": "node",
      "args": ["/path/to/poem-mcp/dist/mcp.js"]
    }
  }
}

Usage

Read a POEM

"Read this poem file and explain what it does"

The poem_read tool parses .poem files into structured elements (constants, structs, functions, enums, etc).

Validate a POEM

"Validate this poem spec"

Checks for syntax errors, naming convention compliance, and structural completeness.

Translate a POEM

"Translate this poem to Python"

The tool parses and validates the spec, then provides a structured translation brief with language-specific type mappings. The LLM then generates idiomatic code.

Supported targets: Python, TypeScript, Go, Rust, Java, Swift, Kotlin, SQL, and any other language.

Example

// pricing.poem
const TAX_RATE = 0.21;

struct Product {
  name: string;
  price: float;
}

fn total(p: Product, qty: int) -> float {
  return p.price * qty * (1 + TAX_RATE);
}

Ask: "Translate to Go" and get:

const TAX_RATE = 0.21

type Product struct {
    Name  string
    Price float64
}

func Total(p Product, qty int64) float64 {
    return p.Price * float64(qty) * (1 + TAX_RATE)
}

Origin

Built by OpenPoem

License

MIT

Available Tools

3 tools
poem_readA

Read and parse a .poem file into structured elements (constants, structs, functions, etc). Returns the parsed structure so you can understand, discuss, or translate the spec.

Use this when someone shares a .poem file or asks about its contents.

Inputs: source (required), filename.

poem_translateA

Prepare a .poem spec for translation to a target language. Parses the POEM, validates it, and returns a structured translation brief.

YOU (the LLM) then use this brief to generate idiomatic code in the target language. The tool handles parsing and validation. You handle the translation.

Supported targets: any programming language (python, typescript, go, rust, java, sql, swift, kotlin, etc).

Inputs: source (required), target (required), style, filename.

poem_validateA

Validate a .poem file for syntax correctness and convention compliance. Checks: brace balance, naming conventions (UPPER_SNAKE for constants, snake_case for functions, PascalCase for types), and structural completeness.

Returns errors (must fix), warnings (should fix), and stats.

Inputs: source (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.2/5.0

    Scored across 3 tools

    Disambiguation5/5

    Each tool targets a distinct operation: read parses and returns structure, validate checks correctness, and translate prepares for conversion. There is no overlap in their purposes, making selection unambiguous.

    Naming Consistency5/5

    All tools follow a consistent 'poem_verb' pattern with clear, specific verbs (read, validate, translate). Naming is predictable and uniform in snake_case.

    Tool Count5/5

    Three tools is well-scoped for a specialized POEM handling server, covering the core needs of parsing, validation, and translation. Each tool earns its place without redundancy.

    Completeness4/5

    The set covers the primary lifecycle of reading, validating, and translating POEM specs. Minor gaps like a create/edit tool are outside the apparent focus, so the surface is mostly complete for its stated purpose.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables coding agents to convert natural language engineering prompts into editable parametric CAD models with deterministic parsing, validation, and edit support.
      5
      Apache 2.0
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables deterministic, neural-free compilation of natural language into typed intent and canonical receipts, with proof-gated synthesis and governed tool use via read-only compiler functions.
      MIT