edobusy/agenthold

Overview

A Python-based MCP server that provides shared versioned state for multi-agent AI workflows. It enables agents to read and write a common state with version control, supporting coordination across multiple AI agents.

Best for

Best for
Developers prototyping multi-agent systems who need simple shared versioned state via MCP.

Use cases

• Orchestrating multi-agent tasks with shared memory
• Tracking conversation history across agent sessions
• Maintaining consistent context in distributed LLM workflows

How to use

Install

pip install agenthold

Tools exposed

• agenthold_register
• agenthold_claim
• agenthold_release
• agenthold_status
• agenthold_wait

Tested with

Claude Desktop, Claude Code, Cursor, Continue, Windsurf, LangChain, LangGraph, CrewAI

Example client config

[object Object]

Notes

A Python-based MCP server that provides shared versioned state for multi-agent AI workflows. It enables agents to read and write a common state with version control, supporting coordination across multiple AI agents.

1 stars on GitHub. Last updated 2026-05-10. Licensed MIT.

Use cases

• Orchestrating multi-agent tasks with shared memory
• Tracking conversation history across agent sessions
• Maintaining consistent context in distributed LLM workflows

Pros

• Open source Python implementation easy to integrate
• Versioned state ensures consistency and rollback capability
• MCP protocol enables interoperability with other MCP tools

Cons

• Very early stage with only 1 GitHub star and limited community
• Requires MCP client support, not standalone
• No documentation or example usage provided yet

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.