# Agiflow

> A shared project board where people and approved AI assistants like ChatGPT, Claude, Cursor, and Codex track tasks, decisions, and next steps via MCP.

Agiflow is a project board for solo builders and small teams who work with AI assistants such as ChatGPT, Claude, Cursor, or Codex. It keeps tasks, decisions, and next steps in one shared workspace that people and approved assistants can read and update through MCP tools. The company also publishes an open-source AI Code Toolkit on GitHub and offers an iPhone app.

## What It Is

Agiflow is a project management board built to be operated alongside external AI assistants. Instead of running its own agents or chat, it provides the board, MCP tools, and scoped access that assistants you already use work through. Tasks move through statuses, and decisions stay attached to the tasks they shaped, so a later session in another tool can pick up where the previous one ended.

## How the Board-Based Workflow Works

The workflow is expressed as project structure rather than a drag-and-drop canvas. Tasks and statuses carry the sequence, work units group related tasks into a scoped chunk of work, and workflow locks prevent two people or two assistants from acting on the same item simultaneously. A human review status lets a person approve work before it moves on. Agiflow states it does not run or host agents.

## Connecting Your Assistants

Assistants connect through an MCP endpoint, and the site shows a one-line Claude Code command for adding it. Docs cover choosing a scope, authorizing access, and verifying ChatGPT, Claude, Cursor, or other assistants. Security documentation covers roles, scoped connections, API keys, storage boundaries, and vault encryption. An iPhone app lets you review projects and update tasks away from the desk.

## Open-Source AI Code Toolkit

The AgiFlow GitHub repository, licensed AGPL-3.0, contains MCP servers for coding agents: scaffold-mcp generates projects and features from templates, architect-mcp supplies file-level design patterns and rule-based review, style-system offers design-system discovery, and one-mcp provides progressive tool discovery. A CLI initializes workspaces and configures MCP for agents including Claude Code, Cursor, Gemini CLI, and Codex CLI.

## Features
- Shared project board for people and AI assistants
- MCP endpoint for assistant access
- Tasks, statuses, and work units
- Workflow locks to prevent concurrent edits
- Human review status
- Decisions attached to tasks
- Scoped connections and API keys
- Vault environments and artifacts
- iPhone app
- Open-source scaffolding, architecture, and design-system MCP servers

## Integrations
ChatGPT, Claude, Claude Code, Cursor, Codex, VS Code, Gemini CLI, GitHub Copilot

## Platforms
LINUX, IOS, WEB, API, VSC_EXTENSION, CLI

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://agiflow.io
- Documentation: https://agiflow.io/docs
- Repository: https://github.com/AgiFlow/aicode-toolkit
- EveryDev.ai: https://www.everydev.ai/tools/agiflow