SwarmForge is an open-source, locally-run orchestration system that lets multiple AI agents collaborate on software engineering projects.
It uses tmux and git to coordinate specialized agent roles (such as coder, architect, and QA), manage separate code branches via worktrees, and pass work between agents through a validated handoff protocol.
Three pre-built workflow templates—two-pack, four-pack, and six-pack—scale from small tasks to major projects, each with different numbers of agent roles and quality gates.
What happened
An open-source project called SwarmForge has been released on GitHub (at github.com/unclebob/swarm-forge) as a tmux-based orchestration platform for coordinating multiple AI agents working together on software engineering tasks. It offers three pre-built workflows (two-pack, four-pack, and six-pack) with 2, 4, and 6 agent roles respectively, each optimized for different project sizes and complexity levels.
Why it matters
Teams using AI agents for development can now coordinate multiple specialized agents (coder, architect, QA, etc.) working in parallel on the same project without stepping on each other's work. SwarmForge handles role assignment, git worktree management, message passing, and backend selection (Claude, Copilot, Codex, or Grok), turning what would be manual coordination into a structured, repeatable workflow.
What to watch
SwarmForge runs entirely locally using tmux and git, requiring only zsh, git, tmux, Babashka, and at least one configured agent backend. Users can start with the quick two-agent workflow, move to the moderate four-agent specification workflow, or use the full six-agent workflow for major projects. The system includes a layered constitution structure that allows project-specific customization while maintaining shared operational scripts.
Ask the AI about this article →
SwarmForge addresses a practical gap in AI agent coordination: while individual agents can write or review code, orchestrating multiple specialized agents without manual handoff work has been cumbersome. The project's three templated workflows (two-pack, four-pack, six-pack) reflect a deliberate design choice to match scope to complexity—small tasks can use a lean two-agent setup, while major projects benefit from full separation of concerns across six roles (specifier, coder, cleaner, architect, hardener, QA). By running entirely in tmux and git on a local machine, SwarmForge avoids external dependencies and keeps all swarm state contained in the project directory (.swarmforge/), making it portable and self-contained. The layered constitution system—where shared default articles live on main and runnable branches can add or override them—allows teams to maintain consistent engineering standards while customizing workflows per project. The validated handoff protocol (enforcing exactly 10-character commit abbreviations and single-line message limits) is designed to prevent silent failures and keep agent communication reliable and auditable.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
Israeli startup DataAgent Ltd
SK Hynix presented a custom HBM concept at SEMICON Taiwan 2026, where compute functions are placed in the base…
Nvidia reported earnings that were both remarkable and boring, reflecting its focus on avoiding a consolidated…
Anthropic has agreed to a $35bn cloud-computing contract with Lambda, a Nvidia-backed cloud provider
The Supreme Court of Japan has included about ¥60 million in its fiscal 2027 budget request for AI-related exp…
The Consumer Affairs Agency said Tuesday it will use generative AI to analyze about 900,000 annual consultatio…
