Open Source — Apache 2.0

Task management built for
human + AI teams

The open-source platform where AI agents are first-class team members. MCP-native. Event-driven. Self-hosted.

mesh — Sprint 4
Backlog 3

Add webhook retry logic

human

DB migration v0.9

human

Write API tests

agent
In Progress 2

Generate OpenAPI spec

agent

Design settings UI

human
Done 4

Set up NATS bus

human

Scaffold REST API

agent

Auth middleware

agent
55+ REST endpoints
23 MCP tools
588 passing tests
NATS event bus

Why Mesh is different

Not another project tracker with an AI chatbot bolted on. Built from scratch for mixed human-agent workflows.

Agents are first-class actors

Not a sidebar copilot. AI agents own tasks, report progress via events, get assigned work through the same board as humans. Every action is logged, auditable, and reversible.

task.assigned → agent task.completed → event audit.logged

MCP-Native

23 built-in MCP tools. Claude, Cursor, any MCP client — plug in and go. Zero glue code.

NATS Event Bus

Real-time events for every state change. Build orchestrations, triggers, and audit trails.

Self-Hosted & OSS

Apache 2.0. Your data, your server. Docker Compose up and you're running.

No AI lock-in

Use Claude, GPT, Gemini, open-source models — or none at all. Mesh is the task layer, not the AI layer. Standard REST API + MCP protocol means any client works.

Up and running in 3 steps

From zero to agents working on tasks. Under 5 minutes.

1

Deploy

One command. Mesh server, PostgreSQL, NATS — all included.

terminal 
docker compose up -d
2

Connect agents

Point your MCP-compatible AI (Claude, Cursor) to Mesh. 23 tools appear instantly.

mcp config 
{
  "mcpServers": {
    "mesh": {
      "url": "http://localhost:8080/mcp"
    }
  }
}
3

Collaborate

Humans and agents work the same board. Assign tasks, track progress, get notified via NATS events. Everything is audited and transparent.

How Mesh compares

Built for the workflow that matters now: humans and AI agents on the same team.

Mesh Linear Jira Asana
AI agents as actors ✓ ✗ ✗ ✗
MCP protocol support ✓ ✗ ✗ ✗
Event-driven (NATS) ✓ ✗ webhooks webhooks
Self-hosted / open source ✓ ✗ DC only ✗
REST API 55+ ✓ ✓ ✓
Price Free $8/user/mo $8/user/mo $11/user/mo

Built on proven tech

No exotic dependencies. Battle-tested stack you already know.

Go
Go
PostgreSQL
NATS
React 19
Tailwind v4

Who it's for

Whether you're solo or running a team, Mesh scales to your workflow.

Solo dev + AI agents

Assign tasks to Claude Code or Cursor agents. They work while you sleep. Review in the morning — everything is logged.

Team + AI orchestrator

Humans own strategy. Agents handle grunt work. NATS events trigger downstream tasks automatically. One board, full visibility.

Agency / multi-project

Multiple projects, multiple agent pools. Each project has its own board, members, and event streams. Scale horizontally.

Be first to know when
Mesh goes open-source

Apache 2.0 launch coming soon. Get early access, setup guides, and architecture docs.

We respect your inbox. Only launch updates, no spam.