SwarmForge is a local tmux-based coordination tool for running multiple AI agents across git worktrees with role-based workflows.
SwarmForge is a Clojure-based open-source tool for coordinating several AI agents on a single project. The README describes it as a disciplined orchestration platform that uses tmux, git worktrees, role-specific prompts, and message passing to help agents collaborate without interfering with one another. It offers different runnable branches for different workflow sizes: two-pack, four-pack, and six-pack.
It addresses the coordination problem that appears when multiple AI agents work on the same codebase at once. According to the README, SwarmForge provides a shared structure for assigning worktrees, separating roles, and managing handoffs so agents can work in parallel without stepping on each other. It is also positioned as a way to keep development disciplined, with explicit review and verification steps depending on the chosen workflow.
Conceptually, SwarmForge works by starting a local swarm from a chosen runnable branch, then launching tmux sessions and branch-specific role prompts for the agents involved. The main branch is documentary and supplies shared operational scripts and default constitution articles; when a runnable branch is started, its wrapper copies those shared pieces locally if needed and then runs the branch’s configuration. Each workflow defines a sequence of agent roles and handoffs, ranging from a compact coder-cleaner loop to broader specification, architecture, hardening, and QA stages.
It appears to be gaining attention because it packages a practical way to run multiple AI coding agents in a structured, local workflow. The README emphasizes disciplined collaboration, clear role separation, and different configurations for small to large projects, which fits current interest in agent orchestration tools. Its weekly star growth also suggests rising visibility, but the README itself does not explain a specific event or release driving the trend.
The README does not name direct competitors. Based on the described approach, the closest comparison is other agent-orchestration or multi-agent coordination setups, especially those that use local terminal automation, tmux, or worktree-based parallelism. SwarmForge’s differentiator in the README is its branching workflow model with distinct role prompts and staged handoffs.
AI-explained · grounded in each repo's README