v2.8.0 — 53 Built-in Tools + 17 Messaging Platforms

Orchestrate AI Agents,
Ship Faster

Squan is a desktop command center that dispatches multiple AI agents to work on your codebase in parallel. Create tasks, chat with agents, and merge their work — all from one interface.

Desktop App
Interactive Agents
Open Source

See It in 90 Seconds

Create a task, dispatch 3 agents, watch them ship a PR.

The Whole Workflow, One App

Kanban board, agent chat, and cost dashboard — all in a native desktop window.

Squan kanban board with tasks flowing from Open to In Progress to PR Review to Landed

Kanban Board

Drag tasks across columns or drive the flow from the sq> console.

Squan agent chat view showing an AI agent responding with tool calls

Agent Chat

Follow-up questions, tool-call breakdowns, mark-complete in one click.

Squan costs dashboard showing per-agent spend and token usage

Cost Dashboard

Live per-agent cost, tokens, and runtime — no surprise bills.

Measurable Impact

Parallel agents, interactive workflows, zero friction

4 Agents in Parallel

Ship 4x faster with concurrent workers

4x
Throughput

Talk to Your Agents

Follow-up questions, even after completion

Interactions

Zero Merge Conflicts

Each agent gets its own git worktree

0
Conflicts

Full Git History

Every task change is a commit

100%
Audit Trail

Everything You Need to Command Your Agents

Six core capabilities that make multi-agent development effortless.

Multi-Agent Orchestration

∞ Agents

Dispatch multiple AI agents in parallel. Each runs in an isolated git worktree — no conflicts, true concurrency.

Interactive Agent Chat

Live

Ask follow-up questions after an agent finishes, give more instructions, or click Mark Complete to advance the board.

Visual Kanban Board

5 Columns

Tasks flow: Open → In Progress → PR Review → Landed. Drag, drop, or drive it from the sq> console.

53 Tools + MCP + Multi-Model

53 Tools

53 built-in tools (filesystem, git, code analysis, network, database, system, agent coordination) + 1000+ MCP servers. Pick any model: Claude, GPT-4o, Gemini, Ollama.

Everything-as-Code

100% Git

All project state lives in .squan/ — tasks as markdown, committed to git. Full audit trail, reviewable in PRs.

Cost & Event Monitoring

Real-time

Live event stream, per-agent cost tracking, metrics dashboard. See exactly what every agent is doing and spending.

How It Works

One desktop app, many agents, one git repo. Each agent gets its own worktree — branches stay clean, merges stay simple.

Squan architecture diagramYou send a task to the Squan Desktop App. The app spawns multiple agents in parallel. Each agent runs in its own isolated git worktree on its own branch. When an agent finishes, it opens a pull request against the main repository. The kanban board tracks task state end-to-end.Youcreate a taskSquan DesktopKanban · ChatCosts · Eventsdispatch →Agent · bee-alphaworktree/bee-alphaAgent · bee-bravoworktree/bee-bravoAgent · bee-charlieworktree/bee-charliePRs→ mainbranch
You dispatch. Agents work in parallel, each in an isolated worktree. PRs converge back to main.

What People Actually Build

Three scenarios where dispatching agents in parallel beats doing it yourself.

Build an MVP in a weekend

Dispatch four agents at once — one on frontend, one on API, one on tests, one on docs. Each works on its own branch. Merge when PRs open.

Shipped in 2 days instead of 2 weeks.

Migrate a legacy codebase

Split the migration by module. Assign one agent per package to upgrade imports, rewrite deprecated APIs, and run the test suite. Review PRs as they land.

30 modules migrated in parallel, not serially.

Backfill test coverage

Point tester-role agents at every untested module. Each writes unit tests, runs them, and opens a PR. You review, merge, and repeat.

From 20% to 80% coverage without a sprint.

Ready to Orchestrate Your Agents?

Download Squan and start dispatching AI agents to build your projects in parallel. Chat with them, give follow-up instructions, and ship faster.