multiwork Download V1.1.0

MULTIWORK · LOCAL AGENT WORKSPACE

Many agents. One workspace.

Bring local CLIs and model APIs into one conversation. Follow each reply, inspect usage and stop individual agents.

One task — across the Agent CLIs you already run.

Agent usage

Sample session
Reported tokens
12,840TOKENS
Reply 01Reply 08
Input10,240
Output2,600
> 0+ Supported Agent CLIs
* 0 Collaboration Modes
* Local Conversations & run history
# 0 Shared task

DETECTED ON YOUR MACHINE

Works with the agents you already run.

multiwork detects a compatible local Agent command and exposes its readiness, capabilities, and session behavior. Approved Agent Connectors can be added through the Agent Connector SDK.

INSIDE AUTO DISCUSSION

Give the task a team.

From a mention to a shared plan. Watch Agents take a code review forward.

Auto Discussion Code Change Review
Task
YOUR TASK

Review this code change. Find risks and suggest improvements.

  1. CodexStarting

    I’ll map the change and bring in two peers to check edge cases and implementation.

  2. CodexPeer request

    @Hermes, check requirements and edge cases. @OpenCodeReview, review the implementation and tests.

  3. HermesEdge cases

    Check empty input and retry behavior so the task can recover from failures.

  4. OpenCodeReviewImplementation review

    The approach looks viable. Add a retry test and check that errors are clear.

  5. CodexNext steps

    Next: cover empty input, add a retry test, and improve error messages for your review.

1 / 5

THE SHORT ANSWER

What is multiwork?

multiwork is a local-first multi-Agent collaboration workspace for research, analysis, writing, planning, review, and implementation. Selected Agents advance the same task through discussion, authorized execution, and peer review. You set the boundaries and make key decisions when needed.

Supported agents 13+

Codex, Claude Code, Gemini CLI, Qwen Code, Kimi Code, MiMo Code, OpenCode, OpenCodeReview, Hermes, Zcode, Pi Agent, OpenClaw, WorkBuddy — detected natively on your machine.

Modes 3

Direct · Concurrent Responses · Auto Discussion

Working together
Discuss → Execute → Check → Refine

Agents build on peer contributions and request the next steps, reducing manual coordination.

Runs on

Windows · x64 · Electron

Local conversations and run history; models may use external providers.

THE WORK CELL

Set up the task. Let the group move it forward.

01

Select

Choose the local Agents and participants. Participants are always selected by you — never auto-assigned from a roster.

02

Scope

Bound the goal, working directory, context, and permissions. Every run starts from a frozen task snapshot.

03

Collaborate

Start Auto Discussion so Agents can discuss, request contributions, carry out authorized work, and check deliverables.

04

Decide & check

Step in for missing information, permissions, or key decisions. Check the deliverables and request changes when needed.

COLLABORATION MODES

Three ways agents work together.

Work with one Agent, compare independent replies, or let a group advance a task together.

>

Direct

One selected Agent keeps its conversation and native session when supported.

Best for focused work with one Agent.
*

Concurrent Responses

Selected Agents receive the same frozen task snapshot and return independent replies in stable order.

Best for comparing approaches before choosing one.
#

Auto Discussion

Agents propose, discuss differences, request peer contributions, carry out authorized work, and check results.

Best for multi-step work with less manual coordination.

INSIDE THE WORK CELL

One workspace for the group and its work.

Codex and Claude Code reviewing a Windows release in multiwork
Windows desktop · Sample group conversation

Follow each participant

In concurrent mode, each reply appears when ready. Stop an individual participant while others continue.

Inspect usage

Expand agent usage to see available token counts. Supported Codex sessions also report quota windows.

Return to a task

Browse local tasks by status and select a card to return to the original request.

Try a task such as reviewing a PR, analyzing a bug, or drafting a plan. Choose compatible Agents, give them shared context, and start Auto Discussion. They can compare approaches, request further work, and check deliverables. Authorize workspace writes when implementation is needed, and respond when they need a key decision.

HOW multiwork DIFFERS

Spend less time managing the handoffs.

multiwork brings the Agent tools you already run into a local-first collaboration workspace. Let the group discuss and advance the task, while you focus on the goal, boundaries, and key decisions.

WHERE multiwork FITS

Where task collaboration fits in your workflow.

multiwork focuses on shared tasks across the local Agent tools you choose. Here is how that focus relates to other ways of organizing Agent work.

Category
What it focuses on
multiwork's difference
Individual Agent sessions
Work with an Agent through its own interface and tools.
Bring selected CLIs into a shared task where Agents can discuss and request peer contributions.
Hosted Agent execution
Run work in a remotely managed environment.
Local Electron desktop app using the CLIs already installed on your machine.
Agent messaging
Exchange messages and events between participants.
Connect discussion to authorized execution, deliverable review, and task progress decisions.
Parallel workspaces
Organize separate tasks or branches running at the same time.
Let different Agent CLIs build on each other's contributions to the same task.
Custom orchestration
Build and maintain your own task routing and collaboration logic.
A ready-to-use desktop workflow across 13+ heterogeneous CLIs — no orchestration framework to assemble.

THE BOUNDARY

What multiwork is, and isn't.

Is

  • A local-first desktop workspace for multi-Agent collaboration.
  • A working group built from the Agent CLIs you select.
  • Shared context, authorized execution, peer review, and local run history.
  • Workspace writes as opt-in workflow controls.
  • Open source under the Apache License 2.0.

Isn't

  • Not a hosted Agent fleet or general Agent framework.
  • Not automatic participant selection from a larger roster.
  • Not a guarantee that every task finishes without your input.
  • Not an operating-system sandbox — writes happen through your shell.
  • Not fully offline — a selected Agent may send prompts to its own CLI provider.

ANSWERS

Questions teams ask before adopting.

Short, honest answers — the long versions live in architecture.md ↗.

Start a conversation

Release notes and SHA-256 ↗
  1. Open the appExtract the whole Windows x64 ZIP and launch multiwork.exe.
  2. Connect an agentSign in to a native CLI and refresh discovery, or configure a compatible model API in agent settings.
  3. Send a shared taskChoose participants and a discussion mode. Expand usage in the conversation to inspect the figures each provider returns.