Loomflow Documentation

Learn how to master the agentic workflow and leverage the full power of Loomflow's AI-driven project management.

AI Assistant Guide

Loomflow AI User Guide

Loomflow provides a powerful AI-driven workflow powered by the Model Context Protocol (MCP). This guide explains how you can use AI assistants (like Claude, Cursor, or Windsurf) to manage your projects effectively.

1 Comparing UI Actions vs. AI Prompts

Core Operations

Interaction Type UI Action (Manual) AI Prompt (Automated)
Switch Project Clicking 'Projects' and selecting from the list. "Set my active project to LOOM."
Create Ticket Click '+ New Ticket', fill out title, type, and description. "Create a task ticket named 'Fix login bug' described as 'Users can't log in with email'."
High Priority Create ticket, then drag it to the top of the stack. "Create a priority ticket for 'Database migration'..."
Update Status Dragging a card between columns or using the dropdown. "Move LOOM-001 to In Progress."
Assign Work Selecting a name from the Assignee dropdown. "Assign LOOM-002 to kirt@example.com."
Set Blocker Selecting 'Blocked' status and choosing a ticket from the list. "Set LOOM-005 as a blocker for LOOM-006."

🕒 Time Tracking (if enabled on project)

Task UI Action AI Prompt Example
Start Timer Open ticket details and click **"Start Timer"**. "Start a timer for ticket PROJ-0042."
Stop Timer with Note Enter a note in the active timer field and click the **Stop icon**. "Stop my timer and add a note: 'Finished the initial draft.'"
Log Time Manually Click **"Log time manually..."** in ticket details, fill the form, and save. "Log 2h 15m for PROJ-0042 with the note 'Pair programming'."
Check Time Spent Look at the **"Total"** label in the Work Log section of a ticket. "How much time has been spent on ticket PROJ-0042?"
List Active Timers View active timers on individual ticket cards or detail pages. "Do I have any timers running right now?"
View Timesheet Click the **Clock icon** in the project navigation bar. (N/A - Use web interface for full reports)

📅 Milestone Management

Task UI Action AI Prompt Example
Create Milestone Navigate to Milestones and click **"New Milestone"**. "Create a new milestone called 'Beta Release' due on March 15th."
List Milestones Navigate to the project dashboard or milestone list. "List all the milestones for this project."
Filter by Milestone Use the **"Milestone"** dropdown in the ticket filter bar. "Show me all the tickets in the 'Q1 Cleanup' milestone."
Update Milestone Edit a milestone from its detail page or the project settings. "Change the description of the 'v1.1' milestone to 'Final polish'."

💡 Smart Estimates

Task UI Action AI Prompt Example
Create with Estimate In the New Ticket form, enter durations like "1d 4h" in the **Estimate** field. "Create a new bug 'Fix login' and set the estimate to 1 day and 4 hours."
Update Estimate Edit the **Estimate** field in ticket details (e.g., change "1h" to "30m"). "Update ticket PROJ-0005 and set the estimate to 30 minutes."
Set Long-term Estimate Enter "2w" or "2 weeks" into the **Estimate** field. "Set the estimate for the 'Refactor Backend' story to 2 weeks."

2 Advanced AI Workflows

The real power of the Loomflow AI comes from chaining multiple actions together in a single prompt.

"The Onboarding Prompt"

"Search for a project related to 'Stripe', set it as my active project, and list all 'To Do' tickets assigned to me."

"The Implementation Start"

"Move LOOM-0034 to In Progress, show me the full description, and then help me draft an implementation plan..."

Best Practices for Prompting

  • • Specify Ticket Numbers: Using LOOM-0001 is always more reliable than "that login ticket".
  • • Establish Context Early: Start your session by setting an active project. This makes subsequent prompts much shorter.
  • • Ask for Guidance: If you're unsure what an AI can do, ask: "What Loomflow tools do you have access to?"

🤖 Pro-Tip: Context Awareness

Loomflow's MCP server is context-aware. If you set an active ticket first, you don't need to repeat the ticket number for subsequent commands:

1. "Set PROJ-0042 as my active ticket."
2. "Start a timer."
3. "Set the estimate to 4 hours."
Advanced Concepts

Agent Implementer Concepts

Agent Implementers are a specialized feature in Loomflow designed to facilitate parallel work between human developers and AI coding agents.

Primary Assignee (User)

The human responsible for the overall completion and quality of the ticket. They receive email notifications and oversee the project.

Agent Implementer (AI)

An AI "worker" assigned to facilitate the specific execution of the task. They have a defined Purpose and Instructions that guide their behavior.

Setting Up an Agent Implementer

1

Define the Agent

Go to your Account Settings page and add a new Agent Implementer with a descriptive Name (e.g., "Documentation Specialist").

2

Define the Purpose

The purpose is a persistent prompt. Example: "You are an expert at technical writing... ensure all documentation is clear..."

3

Assign the Agent

When creating or editing a ticket, select your agent from the Agent Implementer dropdown.

Benefit: Parallel Workflows

By using Agent Implementers, a single project owner can manage a "fleet" of agents specializing in different areas of the codebase.

Refactoring Agent (Tech Debt)
Documentation Agent (Syncing Docs)
Bug Fixer (Queue Management)
Human Owner (Reviewer)
Setup Guide

MCP Client Installation Guide

Loomflow provides a powerful Model Context Protocol (MCP) server that allows AI coding assistants to interact directly with your tickets, projects, and milestones.

High-Level Configuration

Loomflow's MCP server runs as a remote service using the SSE (Server-Sent Events) protocol.

Server URL http://app.loomflow.net:6029/mcp
Authentication Authorization: Bearer <your-token>

How to get your token:

  1. 1. Log in to Loomflow.
  2. 2. Navigate to your Profile Page (click your user icon/name).
  3. 3. Scroll down to the Developer Tools section.
  4. 4. Your active token is displayed there. You can also regenerate it if needed.

1. Cursor

Official Docs ↗

Open Cursor Settings (Cmd + Shift + J), navigate to General > MCP, and click + New MCP Server.

{
  "mcpServers": {
    "loomflow": {
      "url": "http://app.loomflow.net:6029/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>",
        "Content-Type": "application/json"
      }
    }
  }
}

2. Antigravity & Windsurf

Antigravity Docs ↗ Windsurf Docs ↗

Edit the raw mcp_config.json file manually to add the following configuration:

{
  "mcpServers": {
    "loomflow": {
      "serverUrl": "http://app.loomflow.net:6029/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>",
        "Content-Type": "application/json"
      }
    }
  }
}

3. Claude Code

Official Docs ↗

In your terminal, run the following command to add the Loomflow MCP server:

claude mcp add-json loomflow '{"type":"http","url":"http://app.loomflow.net:6029/mcp","headers":{"Authorization":"Bearer <YOUR_TOKEN>"}}'

4. Roo Code

Formerly Cline. Supports MCP via its JSON settings logic.

{
  "loomflow": {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-http", "--url", "http://app.loomflow.net:6029/mcp"],
    "env": { "AUTHORIZATION": "Bearer <YOUR_TOKEN>" }
  }
}

5. GitHub Copilot

Search for github.copilot.chat.mcp.servers in VS Code settings.

  • URL: http://app.loomflow.net:6029/mcp
  • Auth Header: Bearer <YOUR_TOKEN>

6. Claude Desktop

Found in claude_desktop_config.json.

{
  "mcpServers": {
    "loomflow": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-http", "--url", "http://app.loomflow.net:6029/mcp"],
      "env": { "AUTHORIZATION": "Bearer <YOUR_TOKEN>" }
    }
  }
}

7. Gemini CLI

Edit ~/.gemini/settings.json.

{
  "mcpServers": {
    "loomflow": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-http", "--url", "http://app.loomflow.net:6029/mcp"],
      "env": { "AUTHORIZATION": "Bearer <YOUR_TOKEN>" }
    }
  }
}

8. Codex

Set LOOMFLOW_OAUTH_TOKEN env var and add to ~/.codex/config.toml.

[mcp_servers.loomflow]
url = "http://app.loomflow.net:6029/mcp"
bearer_token_env_var = "LOOMFLOW_OAUTH_TOKEN"

9. Slack (Slack MCP Client)

Official Docs ↗

Provide the URL and the Authorization header in your workspace settings.

URL http://app.loomflow.net:6029/mcp
Header Authorization: Bearer <YOUR_TOKEN>