Automation Platform Integrations
Automation Platform Integrations
Operator is not just a standalone TUI — it exposes its ticket orchestration over a REST API and a stdio MCP server, so external automation platforms and agent operating systems can drive it, and Operator can hand work out to them.
This is a distinct integration category from your kanban provider, coding agent, git host, or session wrapper. Those are the building blocks Operator uses; automation platforms are peers that compose with Operator.
The two directions
Operator connects to an automation platform in two complementary directions:
| Direction | What it means | How |
|---|---|---|
| Operator → platform | Export an Operator issuetype/ticket as the platform’s native workflow | operator workflow export --format <fmt> |
| Platform → Operator | The platform calls Operator to create tickets, launch agents, poll the queue | a platform plugin, or Operator’s MCP server |
The portable substrate
Because the “platform → Operator” direction rides on REST + MCP — both portable, widely supported substrates — exposing them once connects Operator to the whole category, not just one tool. Platforms in this space include AGNT.gg, n8n, Activepieces, Windmill, Dify, Flowise, Langflow, OpenAI AgentKit, and Zapier/Make.
Operator already ships:
- a REST API (
operator api, default port7008) with endpoints to create tickets, launch agents, read queue status, export workflows, and raise alerts; - a stdio MCP server (
operator mcp) exposing ~18 orchestration tools, which any MCP-capable client (Claude Code, Cursor, Zed, and MCP-consuming automation platforms) can spawn.
Supported integrations
- AGNT.gg — export Operator workflows
as AGNT graphs, and drive Operator from AGNT workflows via the
operator-plugin.
Write/launch tools mutate your repositories. Only connect platforms you trust, and gate Operator’s MCP write tools with
[mcp].expose_ticket_write_tools.