FAQ
General
Section titled “General”What is DevFlow?
Section titled “What is DevFlow?”DevFlow is an AI Development Workflow Manager. It brings structure, traceability, and quality control to AI-assisted software development. Every piece of work follows a defined lifecycle — from idea through planning and approval to implementation and review — ensuring that humans stay in control while AI agents do the heavy lifting.
Is DevFlow free?
Section titled “Is DevFlow free?”DevFlow offers both personal plans and organization plans.
- Personal plans include Free, Pro, and Max tiers. The Free plan lets you get started at no cost with basic features and limited flows.
- Organization plans include Small, Medium, and Big tiers, scaled to your team size.
Visit the pricing page for current details on features and limits for each plan.
What AI assistants work with DevFlow?
Section titled “What AI assistants work with DevFlow?”DevFlow integrates with AI assistants through the MCP (Model Context Protocol) standard. Any MCP-compatible assistant can connect to DevFlow. The primary integration is with Claude (Anthropic), but any assistant that supports MCP can use DevFlow’s tools to manage flows, create plans, and track progress.
Do I need an AI agent to use DevFlow?
Section titled “Do I need an AI agent to use DevFlow?”No. While DevFlow is designed to work with AI agents, you can manage flows manually. You can create flows, write plans yourself, move them through states, and track time — all without connecting an AI agent. The AI integration enhances the workflow but is not required.
Features
Section titled “Features”How does time tracking work?
Section titled “How does time tracking work?”DevFlow tracks time through work days:
- Timer — Start and stop a timer while you work. It attaches to a flow automatically.
- Manual entries — Log time after the fact by entering start time, end time, and the flow you worked on.
- Calendar view — See your tracked time laid out on a visual calendar.
All time entries are organized by work day, giving you a clear picture of how your time was spent.
What is the Takt system?
Section titled “What is the Takt system?”Takt is DevFlow’s system for scheduled rest break reminders. The organization sets a default interval (for example, 15 minutes), and DevFlow uses it to enforce compliance with labor law regulations. When you have been working continuously and a rest break is required, the timer reminds you to take a break for the configured Takt duration. Individual members can have custom Takt settings if needed. See Compliance for full details.
Organizations
Section titled “Organizations”Can I be in multiple organizations?
Section titled “Can I be in multiple organizations?”Yes. You can belong to any number of organizations and switch between them freely using the context switcher. This is common for freelancers and consultants who work with multiple clients. Each organization has its own projects, settings, and compliance rules, and your role may differ between organizations.
Is my data isolated between personal and organization contexts?
Section titled “Is my data isolated between personal and organization contexts?”Yes, completely. Personal projects are never visible in an organization context, and organization projects are never visible in your personal context. Data from one organization is never accessible from another. This ensures full privacy and prevents accidental data leaks between workspaces.
AI Agent Integration
Section titled “AI Agent Integration”How does the AI agent know my coding standards?
Section titled “How does the AI agent know my coding standards?”DevFlow provides your AI agent with context through two mechanisms:
- Project knowledge base — Each project has a knowledge base where you can store architecture decisions, code conventions, important patterns, and domain-specific information. The agent reads this before starting work.
- Git settings — Projects can configure branch naming conventions, PR templates, and other Git-related settings that the agent follows during implementation.
Together, these ensure the agent writes code that fits your project’s standards.
Can I reject an AI-generated plan?
Section titled “Can I reject an AI-generated plan?”Yes. When the AI agent submits a plan, the flow moves to the approval state. You review the plan and can either approve it or reject it with specific feedback. If you reject, the flow goes back to planning, and the agent revises the plan based on your comments. This loop continues until you are satisfied. You are always in control of what gets implemented.
What happens if the AI agent makes a mistake?
Section titled “What happens if the AI agent makes a mistake?”The review state catches mistakes before they reach production. After the agent finishes implementation, you test the changes using the provided testing instructions and review the code. If something is wrong, you reject the review with feedback, and the agent goes back to fix it. No code is merged without your explicit approval.
How do I track AI agent work time?
Section titled “How do I track AI agent work time?”Agent work time is tracked automatically. When an AI agent works on a flow, DevFlow records agent sessions with start time, end time, and what was accomplished. You can see these sessions in the flow detail view. This gives you visibility into how much time the AI spent on each task, separate from human work time.
Workflow
Section titled “Workflow”What is the difference between a flow and a task?
Section titled “What is the difference between a flow and a task?”A flow is a complete unit of work — a feature, bug fix, or task that moves through the full lifecycle (idea to done). Tasks are sub-items within a flow that break the implementation into smaller, trackable steps. For example, a flow might be “Add user authentication” with tasks like “Set up database schema”, “Create login endpoint”, and “Build login page”.
Can I move a flow backwards?
Section titled “Can I move a flow backwards?”Only at specific points. You can reject a plan during approval (sends it back to planning) and reject an implementation during review (sends it back to in_progress). These rejection loops allow you to course-correct with feedback. You cannot arbitrarily move a flow to any previous state.
What happens when a flow is marked as done?
Section titled “What happens when a flow is marked as done?”When a flow reaches the done state, it is considered complete. The completion timestamp is recorded, the full audit trail is preserved, and the flow appears as finished on the Kanban board and in project statistics. Done flows serve as a permanent record of what was built, who made each decision, and when everything happened.