# Agiflow Documentation

> Give your AI agents project context — connect Claude Code, Cursor, or ChatGPT via MCP

## Introduction

### Welcome to Agiflow

Agiflow bridges the gap between high-level project intent and autonomous AI execution. Connect your AI coding tools via MCP so agents get project plans, tasks, and acceptance criteria — no re-explaining every session. Includes the open-source AI Code Toolkit (MCP servers for scaffolding, architecture, and progressive discovery) and the Agiflow Platform for managing projects, work units, and tasks.

URL: https://agiflow.ai/docs
Keywords: introduction, overview, ai agents, agiflow, MCP, project context, AI Code Toolkit, command layer

## Getting Started

### Quickstart Guide

Sign up, create a project, and connect your AI tool — all free, no credit card required. Add one MCP URL to your AI tool config, authorize via OAuth, and verify by asking your agent to list your projects. Full setup in under 5 minutes.

URL: https://agiflow.ai/docs/getting-started
Keywords: setup, quickstart, tutorial, guide, MCP integration, Claude Code, Cursor, ChatGPT, OAuth, API key

## Connecting AI Tools

### Unified MCP Connection

Connect Claude Code, Cursor, or ChatGPT to Agiflow via a single MCP URL. Choose from 4 scope levels (Organization, Project, Work Unit, Task) for precise access control. Authenticate via OAuth (recommended for interactive tools) or API key (for CI/CD and scripts). Progressive discovery keeps initial token usage low — agents start with 3 meta-tools and discover capabilities on demand.

URL: https://agiflow.ai/docs/connecting-ai-tools
Keywords: MCP integration, Claude Code, Cursor, ChatGPT, OAuth, API key, progressive disclosure, scoped MCP, organization scope, project scope, work unit scope, task scope

## Features

### Workflows

From plan to ship — your AI agent workflow. Plan projects, refine tasks, groom backlogs, execute work units, and triage completed work. MCP skills (prompts) guide each phase: project_plan, refine_task, backlog_grooming, run_work, run_task, triage, review_work, daily_standup. Session continuity via devInfo metadata means agents resume with full context.

URL: https://agiflow.ai/docs/features/workflows
Keywords: workflows, planning, task refinement, backlog grooming, execution, triage, devInfo, session continuity, work units, acceptance criteria

### Security & Governance

Role-Based Access Control (RBAC) with 4 roles (Owner, Admin, Member, Agent). Docker sandbox isolation with restricted file system access, credential mounting, and automatic cleanup. API key permission scoping for CI/CD and agent execution. Upcoming features: per-agent MCP tool proxying, resource quotas, audit logging, network isolation policies.

URL: https://agiflow.ai/docs/features/security
Keywords: security, governance, RBAC, roles, permissions, sandbox, Docker, access control, API keys, isolation

## Open Source MCPs

### Scaffold MCP

Generate standardized code from templates with Liquid syntax. Open-source MCP server for rapid bootstrapping, consistent architecture, and incremental feature addition. Built-in templates: nextjs-15-drizzle, typescript-lib, typescript-mcp-package. Fully extensible for custom templates.

URL: https://agiflow.ai/docs/mcps/scaffold-mcp
Keywords: scaffold-mcp, MCP, scaffolding, templates, boilerplate, code generation, open source, Liquid, AI Code Toolkit

### Architect MCP

Architecture patterns and code validation through architect.yaml and RULES.yaml. Open-source MCP server providing proactive pattern guidance (BEFORE editing) and automated code review (AFTER editing). Severity-rated feedback: CRITICAL, HIGH, MEDIUM, LOW. Template-specific rules for Next.js, TypeScript libraries, and MCP packages.

URL: https://agiflow.ai/docs/mcps/architect-mcp
Keywords: architect-mcp, MCP, architecture, design patterns, validation, code review, rules, open source, AI Code Toolkit

### One MCP

MCP proxy server for progressive tool discovery and reduced token usage. Connect multiple MCP servers through a single proxy that loads tools on-demand, reducing initial token usage by 90%+. Supports stdio, HTTP, and SSE transports with tool blacklisting, environment variable interpolation, and flexible YAML/JSON configuration.

URL: https://agiflow.ai/docs/mcps/one-mcp
Keywords: one-mcp, MCP, proxy, token reduction, progressive discovery, multi-server, aggregation, open source, AI Code Toolkit

## Additional Resources

- GitHub: https://github.com/AgiFlow/aicode-toolkit
- Website: https://agiflow.ai
- Sign Up: https://agiflow.ai/auth
- Community Discussions: https://github.com/AgiFlow/aicode-toolkit/discussions