Sponsored by Deepsite.site

Archrad Deterministic Mcp

Created By
archradhq8 days ago
stdio MCP server from @archrad/deterministic—same engine as the archrad CLI. Validates architecture IR (structural + IR-LINT), merges local PolicyPack YAML, checks export drift vs generated files on disk, and returns static archrad_suggest_fix guidance per built-in rule code. No cloud account required; runs locally in Cursor, Claude Desktop, and other MCP hosts. Apache-2.0. Docs: docs/MCP.md in the repo.
Overview

ArchRad MCP Server — user guide

Audience: Registry listings (e.g. mcp.so), README deep-links, and operators who want a single accurate overview. For transport details, IR size limits, and testing recipes, see MCP.md.

What is ArchRad MCP Server?

ArchRad MCP Server gives AI coding agents (Claude Desktop, Cursor, and other MCP-capable hosts) deterministic architecture governance. When your agent designs or edits a system graph, ArchRad validates the IR (intermediate representation) against structural rules and IR-LINT — before anything is committed.

No LLM in the validation loop. Same engine as the archrad CLI. Apache-2.0.

Installation

npm install -g @archrad/deterministic

The MCP server binary archrad-mcp ships with this package (package.jsonbin).

Alternative (no global install): many hosts accept npx:

{
  "mcpServers": {
    "archrad": {
      "command": "npx",
      "args": ["-y", "--package=@archrad/deterministic", "archrad-mcp"]
    }
  }
}

On Windows, if the command is not on PATH, use the full path to node and to dist/mcp-server.js from a local clone after npm run build — see MCP.md §5.3.

Claude Desktop configuration

{
  "mcpServers": {
    "archrad": {
      "command": "archrad-mcp",
      "args": []
    }
  }
}

MCP tools

ToolPurpose
archrad_validate_irValidate IR JSON against structural rules and built-in IR-LINT, with optional PolicyPack from policiesDirectory. Returns irStructuralFindings, irLintFindings, and a sorted combined list. Curated remediation text for a code is a separate call to archrad_suggest_fix.
archrad_lint_summaryShort summary and counts of findings (same inputs as validate: ir / irPath, optional policiesDirectory).
archrad_validate_driftCompare IR to an on-disk export directory (exportDir) for a given target: python or nodejs only. Surfaces drift-class findings such as DRIFT-MISSING, DRIFT-MODIFIED, DRIFT-EXTRA, and DRIFT-NO-EXPORT, plus lint/structural output from the same pipeline when enabled. Optional policiesDirectory, optional skipIrLint.
archrad_policy_packs_loadCompile / smoke-check PolicyPack YAML or JSON from a directory or in-memory files list. Returns { ok, ruleCount } or errors. Does not register packs for later tools — the process is stateless. To lint with a pack, pass the same path as policiesDirectory on archrad_validate_ir, archrad_lint_summary, or archrad_validate_drift.
archrad_suggest_fixStatic, curated guidance for a built-in rule id (e.g. IR-LINT-MISSING-AUTH-010). Not LLM output; not automatic graph patches.
archrad_list_rule_codesReturns { codes: string[] } — sorted built-in codes that have static guidance in OSS. For descriptions, severity, and compliance language, use archrad_suggest_fix per code or RULE_CODES.md.

Explicit non-goals for OSS MCP are summarized in MCP.md §6–§7.

CLI capabilities (not MCP)

These run in the terminal, not as MCP tools — use them to produce IR before calling archrad_validate_ir:

  • archrad ingest openapi --spec <url-or-file> — OpenAPI 3.x → IR
  • archrad init --from docker-compose.yml — docker-compose → IR
  • archrad ingest backstage --catalog <dir> — Backstage catalog → IR

Full flags: CLI_REFERENCE.md.

Example IR-LINT rules

Sample of built-in codes (canonical catalog: RULE_CODES.md):

RuleDescription
IR-LINT-DIRECT-DB-001Service calls database directly, bypassing service layer
IR-LINT-DIRECT-DB-ACCESS-002HTTP-facing service has direct edge to database
IR-LINT-MISSING-AUTH-010HTTP entry point has no auth boundary
IR-LINT-DEAD-NODE-011Node with no inbound or outbound edges
IR-LINT-MULTIPLE-HTTP-ENTRIES-009Multiple HTTP entry points with no gateway
IR-LINT-NO-HEALTHCHECK-003No health or readiness endpoint defined

Real-world example (Swagger Petstore)

The repo includes a script that clones public YAML via the GitHub tree API, classifies OpenAPI vs blueprint, and validates. Run from the @archrad/deterministic package root after a build:

cd packages/deterministic   # adjust if your clone layout differs
npm run build
node scripts/github-validate-samples.mjs --repo swagger-api/swagger-petstore --max 5

Findings vary by spec revision; treat output as an illustrative run, not a fixed marketing scorecard.

Enterprise — remote MCP with org policies

For teams using ArchRad Cloud, @archrad/remote-mcp is an HTTP MCP gateway that fetches org PolicyPack context from the API using a per-user or per-org bearer token, so agents do not need local policy files. Deployment (e.g. Cloud Run), token issuance, and what is retained for compliance evidence are product-specific — see packages/archrad-remote-mcp/README.md in this monorepo and your ArchRad Cloud documentation. Do not promise a specific audit trail in OSS copy without matching product/legal language.

Server Config

{
  "mcpServers": {
    "archrad": {
      "command": "npx",
      "args": [
        "-y",
        "--package=@archrad/deterministic",
        "archrad-mcp"
      ]
    }
  }
}
Recommend Servers
TraeBuild with Free GPT-4.1 & Claude 3.7. Fully MCP-Ready.
Y GuiA web-based graphical interface for AI chat interactions with support for multiple AI models and MCP (Model Context Protocol) servers.
Serper MCP ServerA Serper MCP Server
CursorThe AI Code Editor
EdgeOne Pages MCPAn MCP service designed for deploying HTML content to EdgeOne Pages and obtaining an accessible public URL.
Visual Studio Code - Open Source ("Code - OSS")Visual Studio Code
Amap Maps高德地图官方 MCP Server
AiimagemultistyleA Model Context Protocol (MCP) server for image generation and manipulation using fal.ai's Stable Diffusion model.
WindsurfThe new purpose-built IDE to harness magic
MiniMax MCPOfficial MiniMax Model Context Protocol (MCP) server that enables interaction with powerful Text to Speech, image generation and video generation APIs.
MCP AdvisorMCP Advisor & Installation - Use the right MCP server for your needs
DeepChatYour AI Partner on Desktop
Jina AI MCP ToolsA Model Context Protocol (MCP) server that integrates with Jina AI Search Foundation APIs.
BlenderBlenderMCP connects Blender to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Blender. This integration enables prompt assisted 3D modeling, scene creation, and manipulation.
RedisA Model Context Protocol server that provides access to Redis databases. This server enables LLMs to interact with Redis key-value stores through a set of standardized tools.
Baidu Map百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Howtocook Mcp基于Anduin2017 / HowToCook (程序员在家做饭指南)的mcp server,帮你推荐菜谱、规划膳食,解决“今天吃什么“的世纪难题; Based on Anduin2017/HowToCook (Programmer's Guide to Cooking at Home), MCP Server helps you recommend recipes, plan meals, and solve the century old problem of "what to eat today"
Tavily Mcp
Zhipu Web SearchZhipu Web Search MCP Server is a search engine specifically designed for large models. It integrates four search engines, allowing users to flexibly compare and switch between them. Building upon the web crawling and ranking capabilities of traditional search engines, it enhances intent recognition capabilities, returning results more suitable for large model processing (such as webpage titles, URLs, summaries, site names, site icons, etc.). This helps AI applications achieve "dynamic knowledge acquisition" and "precise scenario adaptation" capabilities.
ChatWiseThe second fastest AI chatbot™
Playwright McpPlaywright MCP server