Back to Ecosystem Pulse
ToolExperimentalA2AMCPAG-UI
coworker
by Array-Ventures
Open-source agent platform with MCP UI, A2A, app builder, and skills marketplace
TypeScript
Updated Feb 25, 2026
Share:
Summary
Provides an open-source AI agent platform with an MCP UI MCP UI, A2A protocol support A2A protocol, app builder, and a skills marketplace. Composes agents and skills using Mastra and exposes both MCP client/server and A2A channels for multi-agent interactions. Includes multi-provider chat connectors and hooks for deployment via railway and eventing through Inngest.
Why It Matters
As agents increasingly call other agents, platforms that combine A2A protocols with developer tooling make it possible to observe and manage agent interactions. Coworker surfaces the communication layer, UI, and a skills market so teams can instrument agent behavior and compare components. That visibility is useful when building reputation or evaluation layers because it provides the interaction logs and integration points needed for agent-to-agent evaluation and trust signals, including patterns like Capability Discovery Pattern to improve interoperability. Capability Discovery Pattern
Target Use Cases
Developers experimenting with multi-agent apps who want a self-hosted MCP/A2A stack and a marketplace-style workflow. For governance and reliability in multi-agent setups, consider applying the Supervisor Pattern to keep components accountable and manageable.
How It's Used
- Prototyping multi-agent workflows that require MCP UI and A2A messaging
- Testing and iterating on a skills marketplace where agents discover and call specialists
- Connecting multiple LLM providers (OpenAI, Anthropic) with a single self-hosted agent app
Works With
mastraopenaianthropicinngestrailwaygoogle-workspacewhatsapp
Topics
a2a-protocolai-agentai-assistantanthropicapp-builderchatbotelectrongoogle-workspaceinngestmastra+10 more
Similar Tools
openclaw
Keywords
a2amulti-agent trustmcpagent-to-agent evaluation