Skip to content

Releases: MinimalFuture/AgentMesh

0.1.3

23 May 07:22
@zhayujie zhayujie
0.1.3
ab1499b
Compare
Choose a tag to compare
0.1.3 Latest
Latest

We're excited to announce the release of AgentMesh 0.3.0 with several significant improvements and new features.

What's New

  • Terminal Tool Integration: Added a built-in terminal tool that enables agents to access file systems and other computer resources through command line. This feature is already supported in the default Agent Team templates.

  • SDK Upgrade: agentmesh-sdk has been upgraded to version 0.3.0, with improved streaming support for multi-agent execution processes, making it easier to integrate with various channels. Several known issues have also been fixed.

    pip install --upgrade agentmesh-sdk==0.3.0

  • Integration with chatgpt-on-wechat: AgentMesh now integrates as a plugin with the chatgpt-on-wechat project, enabling multi-agent capabilities across all messaging platforms supported by the cow project. For detailed infomation, please refer to agent plugin

  • Claude 4 Sonnet Support: Added support for Claude 4 Sonnet model. You can use it by setting the model field to claude-sonnet-4-0 or claude-sonnet-4-20250514.

  • Bug Fix: Resolved browser-use version compatibility issues.

Feedback and Contributions

We welcome your feedback and contributions! Please:

  • ⭐️ Star the project to receive notifications about updates
  • Submit PRs to contribute to the project
  • Open issues for bugs or feature requests

Thank you for your interest in AgentMesh!

Assets 2
Loading

0.1.0

25 Apr 08:02
@zhayujie zhayujie
0.1.0
9f72bb2
Compare
Choose a tag to compare
0.1.0

AgentMesh 0.1.0

We're excited to announce the first official release of AgentMesh, a powerful Multi-Agent platform for AI agents building. AgentMesh provides a framework for inter-agent communication, task planning, and autonomous decision-making, allowing you to build agent teams quickly and solve complex tasks through agent collaboration.

Key Features

Agent Collaboration

  • Zero-code Agent Configuration: Define specialized agent roles with custom system prompts through simple YAML configuration files - no coding required
  • Autonomous Decision-Making: Agents can decide which team member should handle specific subtasks
  • Team-based Task Execution: Support for complex task decomposition with context sharing between team members

Multi-Modal Models

  • Unified API: Seamless integration with leading LLMs including:
    • OpenAI (GPT-4.1, GPT-4o, GPT-4.1-mini, etc.)
    • Claude (Claude-3-7-sonnet-latest, etc.)
    • DeepSeek (DeepSeek-chat, etc.)
    • More models coming soon

Extensible Tools

  • Built-in Tools:
    • Calculator for mathematical operations
    • Current Time for accurate time awareness
    • Browser for web access and interaction
    • Google Search for retrieving up-to-date information
    • File System for saving files
  • Tool Framework: Easily extend with custom tools

Multi-Platform Support

  • CLI: Run via command line interface
  • Docker: Containerized deployment
  • SDK: Programmatic integration into your applications

What's Next

We're actively working on enhancing AgentMesh with:

  • WebUI: A user-friendly interface for managing agent teams
  • MCP Protocol: Extended tool capabilities through the MCP protocol
  • Remote Agent Communication: Protocol for heterogeneous agent collaboration
  • More Models: Support for additional LLMs including local open-source models

Feedback and Contributions

We welcome your feedback and contributions! Please:

  • ⭐️ Star the project to receive notifications about updates
  • Submit PRs to contribute to the project
  • Open issues for bugs or feature requests

Thank you for your interest in AgentMesh!

Loading