Kaiban Distributed

Kaiban Distributed is a community project that wraps KaibanJS in a production-oriented, multi-process runtime: workers consume tasks from queues, a gateway exposes HTTP APIs, and state can stream to browsers for a live board experience.

• Repository: github.com/andreibesleaga/kaiban-distributed
• Author: @andreibesleaga
• Integration guide (upstream): KAIBANJS_INTEGRATION.md

Relationship to KaibanJS

KaibanJS provides agents, teams, tasks, tools, and the Kanban-style state model. Kaiban Distributed adds scalable execution across nodes (e.g. BullMQ on Redis or Kafka), retries and dead-letter handling, a gateway with JSON-RPC 2.0 task APIs, optional MCP federation, security hooks (firewall, circuit breaker, JIT tokens, TLS), and OpenTelemetry hooks. Use it when a single Node process is not enough.

When to consider it

• You want several worker processes or containers, each running KaibanJS agents, coordinated by messages.
• You need durable queues, retries, and failure visibility (e.g. completion / DLQ topics).
• You want a central HTTP API to submit work and optional real-time board updates over Socket.io with Redis pub/sub.
• You are aligning with patterns described in our docs for A2A, MCP, Kaiban Board, and telemetry—Kaiban Distributed is a separate codebase that implements its own gateway and observability hooks.

Integration modes (summary)

The upstream guide describes two main patterns:

1. Distributed agent bridge — createKaibanTaskHandler turns a KaibanJS Agent into a queue-backed handler run by an AgentActor (one agent per worker role).
2. Team bridge — KaibanTeamBridge runs a full KaibanJS Team in one process and publishes Zustand state deltas for remote viewers.

For message schemas, Docker examples (including a multi-agent blog pipeline), HITL orchestration patterns, WorkflowDrivenAgent on workers, and environment variables, follow KAIBANJS_INTEGRATION.md—that document is the canonical reference.

Related KaibanJS documentation

• Deploying Your Kaiban Board — static board hosting; Kaiban Distributed adds a distributed live-view pattern.
• A2A Protocol Integration — Google A2A patterns in KaibanJS; Kaiban Distributed exposes its own JSON-RPC gateway for cross-service task submission.
• MCP Adapter Integration — MCP in KaibanJS; Kaiban Distributed includes MCP federation for tools at the infrastructure layer.
• Exporting Traces with OpenTelemetry — tracing concepts; Kaiban Distributed documents OTLP configuration in its integration guide.

Support

Questions, bugs, and contributions belong in andreibesleaga/kaiban-distributed. The KaibanJS team may list or link to community projects here without taking ownership of the external codebase.