Molt

A Claude Code workflow kit that sheds its shell as the way we work with Claude changes.

Today: orchestration skills for Claude Code.

Molt bundles the skills I use to work with Claude Code, and changes as that way of working does. Today that means one orchestrator session that runs related tasks in parallel sessions, tracks them on a live plan board and keeps a private wiki so it can run for weeks. It also lets cloud sessions use your computer. Formerly Cloud Local Bridge: if you installed it under that name, remove it and add Molt.

Install free See what it does
01 ONE CHAT

You describe the work to one orchestrator.

02 PARALLEL SESSIONS

Each task runs in its own session and branch.

03 ONE PLACE

Questions, progress and pull requests come back to you.

What is a Claude Code orchestrator?

A Claude Code orchestrator is one session that plans the work and hands each task to its own session, instead of doing everything in one long chat. You talk only to the orchestrator. It starts the sessions, passes their questions to you, checks each pull request and tracks progress. Molt gives Claude Code the skills to work this way. The orchestrator runs in a claude.ai/code cloud session, and its tasks run in cloud sessions or on your computer. How it works.

The skills in the kit

Four Claude Code skills. Claude picks the right one from what you ask.

orchestrate

Splits a request into tasks and runs each in its own Claude Code session, in the cloud or on your computer. You talk to one session; it relays questions, checks pull requests and asks before archiving finished sessions.

artifact-plan

A live plan board for work with several dependent tasks. It shows every task, its order and what blocks it, and updates as each session reports. A task only moves to done once its pull request is checked.

orchestrator-wiki

A private wiki for a long-running orchestrator, one per domain. It saves decisions and running work as they happen, so you can /compact or /clear and say "resume my billing orchestrator". Old items move to a monthly archive, and the orchestrator reviews its own friction to learn rules.

cloud-local-bridge

Hands browser use, OAuth sign-ins, local files, desktop apps and local MCP servers from a claude.ai/code cloud session to a Claude session on your computer, and sends the result back.

Everything it does

Pick a feature to watch it run.

      Install once

      Add it to your claude.ai account. Cloud and local sessions both get it.

      1. Open Customize → Plugins.
      2. Find Molt under Discover. Choose Add.
      3. Start a new session.
      Or in a local Claude Code session
      /plugin install molt --marketplace ShiJianXi/molt

      Free and open source. MIT license. No data collected.

      Questions

      How do I orchestrate several Claude Code sessions at once?

      Install Molt and say "orchestrate these tasks, one session each". The orchestrator starts a session per task, each on its own branch, and passes their questions to you in one place. Say "plan this and track it on a board" to get a live plan board.

      How is this different from subagents?

      A subagent runs inside one session and reports back once. Molt's orchestrator starts full sessions, each on its own branch, that can ask you questions while they work. Code tasks usually end in their own pull request, and for bigger plans with dependent tasks it offers a live plan board.

      How do I let Claude Code on the web do something that needs my computer?

      Ask it to hand that step to your computer. A local Claude session does it and sends the result back. Browser use, OAuth sign-ins, local files and desktop apps all work this way.

      Why does SendMessage say "No agent named 'session_…' is reachable"?

      Local sessions can't reach a cloud session by its id, only by its title. The plugin uses the title for you.

      Which local sessions work?

      Claude Desktop, or a terminal session started with claude remote-control, on the same account.

      Can the cloud session start a local session for me?

      Yes, while claude remote-control --spawn worktree runs on your computer. Each new session gets its own git worktree.

      Does it collect data?

      No. No servers, no hooks, no network calls. It's instructions for Claude.

      Can I clear or compact a long-running orchestrator?

      Yes, if it keeps a wiki. It saves decisions and what's running as they happen, so /compact and /clear lose nothing: prefer /compact, which usually keeps it working from the wiki; after /clear, say "resume my billing orchestrator" and it reads the wiki first. Each wiki is private to the session that owns it.

      Won't a wiki kept for months get huge?

      No. The pages it reads on every resume stay capped. Old items move to a monthly archive in the same private wiki, and nothing is deleted. About once a week the orchestrator reviews its own friction: it tidies up and adds its own learned rules, and asks you before changing your rules or the scope. The review never runs in another session or on a schedule.

      Why doesn't a session started by another session have my plugins?

      Sessions you start from the app get your claude.ai plugins. A session another session starts can come up with only the built-in ones. The orchestrator tells each child which plugins it uses and where they live, and the child loads their skills itself. Hooks and MCP servers don't come across. A plugin step that must happen goes in the child's done-when, and the orchestrator checks it before calling the task done. Attaching a plugin's repository needs your yes, so the orchestrator asks once before it starts the children.

      Is it free?

      Yes. Free and open source, MIT license.