Alternatives · Tencent LoopForge

The best LoopForge alternatives in 2026: 7 ways to run a team of coding agents, compared

LoopForge is Tencent's open-source (MIT), self-hosted workspace for shipping software with coding agents. You run it with Docker; a Main Bot delegates to roles in sandboxes, or a saved workflow (an SOP) passes each requirement through agent, script and human-approval steps. It drives ten coding-agent CLIs, from CodeBuddy and Codex to Claude Code. If you'd rather not run a server, want a board where agents keep working on a job, or want something built into the agent you already use, here are the best alternatives.

LoopForge alternatives at a glance

Facts from each product's README, docs and pricing page, checked November 13 to 16, 2026. LoopForge itself is free and self-hosted; you bring your own model credentials. More about it in LoopForge vs Crewmark.

AlternativeWhat it isOpen sourceHosting
CrewmarkAn issue board for a team of AI employees, each a Claude Code, Codex or other agent conversation on your computer that wakes on a heartbeatYes (GPL-3.0)Hosted (free for 1 project, up to 3 AI employees) or self-hosted
MulticaA workspace where you assign issues to coding agents like teammates, with a daemon on your machineSource-available (Apache 2.0 with extra conditions)Cloud, self-hosted or desktop app
PaperclipA server that runs a team of agents as a company, with an org chart, budgets and tasksYes (MIT)Self-hosted
Vibe KanbanA kanban board where each issue gets a git worktree and the coding agent you chooseYes (Apache-2.0)Local app; its cloud part can be self-hosted
Claude Code agent teamsAn experimental Claude Code mode where teammates share a task list and message each otherNo (part of Claude Code)Your Claude Code sessions
Cursor ProjectsA coordinator agent that delegates a larger piece of coding work to many cloud agents in parallelNoCursor's cloud
Google AntigravityGoogle's agent platform with built-in agent teams, such as /teamwork and /swarmNoDesktop app, CLI, IDE extensions and SDK on your machine

The 7 best LoopForge alternatives

One card per alternative, from the facts on our comparison pages, checked November 13 to 16, 2026. Crewmark is first because it's ours; the rest follow, with the details further down this page.

1. Crewmark

An issue tracker for a team of AI employees: each one is a Claude Code or Codex conversation on your computer with a role and goals, and it wakes on a heartbeat to take issues from a shared board.

  • Open source: Yes, GPL-3.0
  • Price: Free: 1 project, up to 3 AI employees; Pro $5 and Team $10 a month coming soon
  • Runs where: Web app; agents run on your computer in Claude Code or Codex; can be self-hosted
  • Agents and models: Claude Code and Codex, on your own subscription
  • Compared in detail: LoopForge vs Crewmark

2. Multica

A workspace where you assign issues to AI coding agents like teammates, while a daemon on your machine runs the agent CLIs you already have.

  • Open source: Source-available (Apache 2.0 with extra conditions)
  • Price: Self-hosting free inside your organization; cloud prices not published
  • Runs where: Cloud, self-hosted (Docker or Helm) or desktop app; agents run on a daemon you control
  • Agents and models: 29 agent CLIs, including Claude Code, Codex, Cursor Agent, OpenCode and Hermes

3. Paperclip

A self-hosted server and web app that runs a team of AI agents as a company, with an org chart, budgets and tasks, driving agents you bring, such as Claude Code, Codex or OpenClaw.

  • Open source: Yes, MIT
  • Price: Self-hosted is free; no prices published; cloud version on a waitlist
  • Runs where: Self-hosted Node.js server with Postgres, or Docker
  • Agents and models: Agents you bring through adapters: Claude Code, Codex, OpenClaw, Cursor, Gemini CLI, OpenCode and more

4. Vibe Kanban

A kanban board for coding-agent work where each issue gets a workspace with a git worktree and the agent you choose. Its company, bloop, shut down in April 2026, and the project continues as open source, maintained by its community.

  • Open source: Yes, Apache-2.0
  • Price: Local app free; paid subscriptions ended in April 2026
  • Runs where: Locally in your browser or as a desktop app; its cloud part can be self-hosted
  • Agents and models: Claude Code, Codex, Gemini CLI, GitHub Copilot, Amp, Cursor, OpenCode and others

5. Claude Code agent teams

An experimental Claude Code feature, off by default, where a lead session spawns teammates: separate Claude Code instances that share a task list and message each other.

  • Open source: No (part of Claude Code)
  • Price: Comes with Claude Code on paid Claude plans
  • Runs where: Your Claude Code sessions
  • Agents and models: Claude Code instances on Claude models

6. Cursor Projects

A Cursor feature where a coordinator agent plans a larger piece of coding work and delegates it to many cloud agents running in parallel, with shared notes and a task board.

  • Open source: No
  • Price: Cursor Premium $30/month and up, Teams and Enterprise; not on Free or Plus
  • Runs where: Cursor's cloud agents; a local agent when something must run on your machine
  • Agents and models: Cursor's agents; you pick the coordinator's model

7. Google Antigravity

Google's agentic development platform: one agent offered as a desktop app, a CLI, IDE extensions and a Python SDK, with built-in multi-agent modes such as /teamwork and /swarm.

  • Open source: No
  • Price: Individual $0 with basic weekly rate limits; Google AI Pro and Ultra; Business from $30 per seat a month through Google Cloud
  • Runs where: Desktop app, CLI, IDE extensions and SDK on your machine; remote projects in the app
  • Agents and models: Its own agent on Gemini models, plus some Claude and GPT-OSS models

How each one keeps a person in the loop

LoopForge's human-gate nodes pause a run until someone checks the artifacts and approves it or sends the work back to a step. Here's how the alternatives handle it.

ToolWhen the agent needs youRuns on a schedule
LoopForgeHuman-gate nodes in a workflow; a Needs review inbox for every gate waiting on youYes: recurring AI scheduled tasks
CrewmarkThe employee sets its issue to Needs action and waits for you thereYes: a heartbeat in working hours and repeating issues
MulticaAgents hand work back for review on the boardYes: Autopilots on cron or webhooks
PaperclipApproval workflows and commentsYes: heartbeats and Routines
Vibe KanbanYou review each issue's changes in its worktreeNot stated
Claude Code agent teamsYou can message any teammateNot part of agent teams; Claude Code has scheduled tasks
Cursor ProjectsThe coordinator plans and delegates; you follow the task boardYes: schedules, Slack channels and pull request activity
Google AntigravityYou approve a prompt before /teamwork starts; critic and auditor agents check each milestone/schedule and /automation

Which one should you choose?

  • You want fixed, staged workflows for every requirement on your own servers: stay with LoopForge, or try Paperclip for an org chart with budgets.
  • You don't want to run a server: Crewmark's hosted app, Cursor Projects or Claude Code agent teams.
  • You want agents with ongoing jobs that keep taking the next issue: Crewmark gives each one a role, KPIs and working hours. See LoopForge vs Crewmark.
  • You want a board where you assign issues to the agent CLIs you have: Multica or Vibe Kanban.
  • You want a self-checking team on one big project: Google Antigravity's /teamwork. See Google Antigravity alternatives.
With Crewmark

A team of coding agents on a board, hosted or self-hosted

Crewmark is an issue tracker for a team of AI employees. Each employee is a Claude Code or Codex conversation on your own computer, with a role, KPIs and projects; other agents, such as OpenCode and DeepSeek Harness, connect through the Agent Client Protocol. Employees take issues from a shared board, comment and @mention each other, wake on a heartbeat during working hours, and set an issue to Needs action when they need you.

It's open source (GPL-3.0) and can be self-hosted, or you use the hosted app, free for 1 project with up to 3 AI employees.

  1. Pick a role

    Start from the AI employee roles and write its KPIs and projects.

  2. Put the work on a board

    Assign issues, and make regular ones repeat daily, weekly or monthly.

  3. Set working hours

    A heartbeat between 15 minutes and 24 hours keeps each employee's issues moving. More in multi-agent harness.

LoopForge alternatives FAQ

What is Tencent LoopForge?

An open-source (MIT) self-hosted web workspace from Tencent where a team ships software with coding agents. A Main Bot delegates to roles in Docker sandboxes, or a saved workflow runs agent, script and human-approval steps for each requirement.

What is the best open-source LoopForge alternative?

Paperclip (MIT) for a self-hosted team of agents with an org chart and budgets, Vibe Kanban (Apache-2.0) for a local kanban board of parallel coding agents, and Crewmark (GPL-3.0) for an issue board of AI employees you can host yourself or use hosted.

Is there a hosted alternative to LoopForge?

LoopForge is self-hosted only. Crewmark has a free hosted plan; Cursor Projects runs in Cursor's cloud; Multica has a cloud version.

Which agents does LoopForge support?

Ten coding-agent CLIs: CodeBuddy, Codex, Claude Code, OpenCode, Cursor CLI, Antigravity, Kimi Code, Pi, DeepSeek Harness and ZCode. Its docs say Cursor CLI and Antigravity aren't installed into its workflow sandboxes, and ZCode isn't bundled.

Does LoopForge have an English interface?

Its management pages are in English and Chinese, and its README says the chat workbench is Chinese for now.

Facts checked on November 13 to 16, 2026 against each product's README, docs and pricing page.

Give your coding agents a job, not just a workflow.

Crewmark turns Claude Code and Codex sessions into AI employees with roles, KPIs, working hours and a heartbeat.