Skip to main content
dotbot gives teams a structured way to use AI in software development. Rather than generating code in a black box, dotbot runs each task through a transparent, version-controlled pipeline — so your teammates can see what was decided, why, and how to pick up where it left off. Every AI session, question, answer, and code change is tracked in your repo.

​
What problem does dotbot solve?

Most AI coding tools produce output without a record of how they got there. There is no trail of decisions for teammates to review, no way to resume work across sessions, and no structure for managing large projects with multiple in-flight tasks. dotbot solves this by wrapping AI execution in a managed workflow. Each task passes through a two-phase process (analysis, then implementation), runs in its own isolated git branch, and commits with structured metadata. The result is an AI-augmented development process that is auditable, resumable, and safe to run in parallel across a team.

​
Key capabilities

  • Workflow pipelines — Define multi-step pipelines in workflow.yaml manifests. A project can have multiple workflows installed and run independently. Tasks can be prompt (AI-executed), script (PowerShell, no LLM), mcp (tool call), or task_gen (dynamically generated sub-tasks).
  • Two-phase execution — Analysis resolves ambiguity, identifies affected files, and builds a context package. Implementation consumes that package and writes code. Task state flows: todo → analysing → analysed → in-progress → done.
  • Git worktree isolation — Each task runs in its own worktree on an isolated branch and is squash-merged back to main on completion. Parallel tasks never conflict.
  • Multi-provider AI — Switch between Claude, Codex, and Gemini from the Settings tab. Per-task model overrides let you use cheaper models where they suffice.
  • MCP server — A pure PowerShell MCP server exposes 33 tools for task management, decision tracking, session management, and more. Tools are auto-discovered — no registration required.
  • Web dashboard — A seven-tab UI (Overview, Product, Roadmap, Processes, Decisions, Workflow, and Settings) gives you a real-time view of all running workflows, task queues, and session history.
  • Zero dependencies — The MCP server and web UI are pure PowerShell. No npm, pip, or Docker required. Runs on Windows, macOS, and Linux.

​
Prerequisites

Before installing dotbot, make sure you have the following:
RequirementMinimum versionDownload
PowerShell7.0+aka.ms/powershell
GitAnygit-scm.com/downloads
AI CLI (at least one)LatestClaude CLI, Codex CLI, or Gemini CLI
You can verify your PowerShell version by running $PSVersionTable.PSVersion in a terminal.

​
Get started

Quickstart

Install dotbot, initialize your project, and run your first workflow in minutes.

Installation

All install methods: one-liner, PowerShell Gallery, Scoop, Homebrew, and git clone.

Workflows and stacks

Learn how workflows define pipelines and stacks add tech-specific capabilities.

MCP server

Explore the 33 MCP tools and how to connect dotbot to Claude, Warp, and other AI tools.