MCP.so
ログイン

Marchward

@marchward

リポジトリ

Marchward について

Marchward is an Apache-2.0 open-source runtime authority for AI agents. It routes an agent's tool calls through one control point: credential mediation (the agent holds a single Marchward key and never your real downstream credentials), an inference spend cap enforced before the

#ai-agents#agent-governance#runtime-authority#credential-mediation#mcp#security#audit-log#cost-control#approval-gates#langgraph

基本情報

カテゴリ

AI とエージェント

トランスポート

stdio

公開者

marchward

投稿者

Team Marchward

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "marchward": {
      "command": "npx",
      "args": [
        "-y",
        "@marchward/mcp-server"
      ],
      "env": {
        "MARCHWARD_API_URL": "https://api.marchward.ai",
        "MARCHWARD_API_KEY": "mw_your_key_here"
      }
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Marchward?

Marchward is an Apache-2.0 open-source runtime authority for AI agents. It routes an agent's tool calls through a single control point, providing credential mediation, inference spend capping, human-in-the-loop approval gates, and a tamper-evident hash-chained audit log. It is available as npm packages (@marchward/*), a PyPI package (marchward), and an official MCP registry server (ai.marchward/mcp-server), with a runnable governed-LangGraph example.

How to use Marchward?

Install the Marchward package from npm, PyPI, or use the official MCP registry server. A runnable governed-LangGraph example is provided in the repository to get started.

Key features of Marchward

  • Credential mediation: agent uses a single Marchward key, never real downstream credentials
  • Inference spend cap enforced before calls reach the provider
  • Human-in-the-loop approval gates for irreversible actions
  • Tamper-evident hash-chained audit log for all tool calls
  • Open-source under Apache-2.0 license

Use cases of Marchward

  • Control tool calls made by autonomous AI agents
  • Prevent leakage of sensitive downstream API credentials
  • Enforce budget limits on agent inference costs
  • Require human approval before destructive or irreversible actions
  • Maintain a verifiable audit trail of all agent tool invocations

FAQ from Marchward

How does Marchward mediate credentials?

The agent holds a single Marchward key and never the user’s real downstream credentials. Marchward routes and mediates all tool calls, keeping downstream secrets hidden from the agent.

Does Marchward support human approval?

Yes, Marchward implements human-in-the-loop approval gates that must be passed before irreversible actions are executed.

How does Marchward control spending?

Marchward enforces an inference spend cap before the tool call reaches the provider, preventing budget overruns.

What license is Marchward released under?

Marchward is released under the Apache-2.0 open-source license.

Where can I get Marchward?

Marchward is available on npm (@marchward/*), PyPI (marchward), and the official MCP registry (ai.marchward/mcp-server). A governed-LangGraph example is included in the public repository.

コメント

「AI とエージェント」の他のコンテンツ

LinkedIn MCP Server

stickerdaniel

Open-source MCP server for LinkedIn. Give Claude and any MCP-compatible AI agent access to profiles, companies, jobs, and messages.

2.7K

meGPT - upload an author's content into an LLM

adrianco

Code to process many kinds of content by an author into an MCP server

287

Just Prompt - A lightweight MCP server for LLM providers

disler

just-prompt is an MCP server that provides a unified interface to top LLM providers (OpenAI, Anthropic, Google Gemini, Groq, DeepSeek, and Ollama)

733

Open Multi-Agent Canvas

CopilotKit

The open-source multi-agent chat interface that lets you manage multiple agents in one dynamic conversation and add MCP servers for deep research

507

MCP Client for Ollama (ollmcp)

jonigl

Harness the power of local LLMs with this TUI MCP Client for Ollama. Featuring all core MCP primitives (tools, prompts, resources), agent mode, multi-server, model switching, streaming responses, human-in-the-loop, thinking mode, model params config, system prompts, and saved pre

768