
Multica
Project management for teams of humans and coding agents
About Multica
Multica treats coding agents as members of a team rather than as a tool you invoke. You assign tasks, watch progress, and manage humans and agents in the same place, which is a different proposition from running one agent in a terminal and hoping it finishes. Its own framing is blunt: your next ten hires will not be human.
The practical hook is breadth. Multica supports 23 coding tools out of the box, including Claude Code, Codex, Cursor, Copilot, OpenCode, OpenClaw, Gemini and Kimi, and a daemon auto-detects whichever CLIs you already have installed and registers a runtime for each. That matters because most teams have not standardised on one agent and are not going to; the coordination layer has to be agnostic or it is useless. It runs self-hosted through Docker Compose or Kubernetes, or as a hosted cloud service, and there are no artificial caps on agent count in the self-hosted version.
One thing to be clear about, because the site is not. Multica calls itself "fully open source", and the licence is Apache 2.0 with additional conditions rather than plain Apache 2.0. Those conditions are worth reading before you build on it: you may not offer Multica as a hosted service to third parties without a commercial licence, and that restriction applies even if you charge nothing, and you may not remove the Multica branding without a written waiver. Internal use inside a single organisation, including multiple workspaces, is explicitly fine. So for the common case, running it for your own team, it is free and the source is yours to inspect. For anyone planning to wrap it in a product, it is not.
You deploy it yourself with Docker Compose or Kubernetes, or use the hosted version. A daemon scans the machine for coding CLIs you already have and registers a runtime for each one it finds, so onboarding an agent is largely a matter of already having it installed. From there you create tasks and assign them to agents or people, the same way you would in any tracker, and progress is reported back into one board rather than living in separate terminal sessions. Agents accumulate skills over time, which is the part that distinguishes it from a job queue: the intent is that a given agent gets better at the work you keep giving it. Multiple machines can be connected as runtimes, and each agent has configurable concurrency limits, so throughput is bounded by your hardware rather than a plan tier.
- •23 Coding Tools Supported - Claude Code, Codex, Cursor, Copilot, OpenCode, OpenClaw, Gemini, Kimi and more, auto-detected from what is already installed
- •Humans and Agents on One Board - Tasks are assigned and tracked in the same place regardless of who or what is doing them
- •Compounding Skills - Agents accumulate capability on repeated work rather than starting cold each run
- •Self-Host or Cloud - Docker Compose or Kubernetes on your own infrastructure, or a hosted version, with the same feature set
- •No Artificial Caps - Agent concurrency is limited by your hardware, and multiple machines can be joined as runtimes
- •Source Available - Every line is inspectable and forkable, under Apache 2.0 plus the conditions described above
Teams already running more than one coding agent who have lost track of what each is doing, and anyone whose agents live in separate terminals with no shared view of progress. The breadth of CLI support makes it a reasonable bet if your team has not converged on a single agent, which most have not. It suits organisations that want this on their own infrastructure, since self-hosting is a first-class path rather than an afterthought. Read the licence first if you intend to offer it to anyone outside your own organisation, because that specific case needs a commercial licence even when you give it away for free. If you run a single agent occasionally, this is a coordination layer for a problem you do not yet have.
Pricing
No pricing page found
Free to self-host with Docker Compose or Kubernetes, no caps on agents. A hosted cloud version offers a free trial but publishes no prices: multica.ai/pricing returns 404 as of 2026-08-26.
Checked 2026-09-18













