Skip to main content
Glama
flow-master-io

@flowmaster-io/mcp

@flowmaster-io/mcp

Connect an AI assistant to your Flow Master workspace.

Once configured, assistants like Kiro, Claude Desktop and Cursor can look up your projects, open and update tasks and tickets, move work along your boards, link related items, and read or write comments — using your own Flow Master account and permissions.

The server runs locally on your machine and talks to your Flow Master installation over HTTPS.

Requirements

  • Node.js 20 or newer

  • A Flow Master account

  • Your workspace URLs, available from your Flow Master administrator

Related MCP server: toflow MCP Server

Setup

Add the server to your MCP client's configuration:

{
  "mcpServers": {
    "flowmaster": {
      "command": "npx",
      "args": ["-y", "@flowmaster-io/mcp"],
      "env": {
        "FLOWMASTER_API_URL": "https://api.example.com",
        "FLOWMASTER_WEB_URL": "https://app.example.com"
      }
    }
  }
}

Replace both URLs with the ones for your workspace.

Then sign in once:

FLOWMASTER_WEB_URL=https://app.example.com npx -y @flowmaster-io/mcp login

Your browser opens, you sign in the way you normally do, and the session is stored at ~/.flowmaster/mcp.json with owner-only permissions. Restart your MCP client afterwards.

What the assistant can do

Projects and tasks:

Tool

What it does

list_projects

Find a project by name or code

get_project

Read one project by id

list_tasks

Read the tasks in one project, with their descriptions and tags

get_task

Read one task by id

create_task

Open a task in a project

update_task

Change a task's title, description, priority, due date or tags

Tickets:

Tool

What it does

list_tickets

Read tickets with their descriptions and tags

get_ticket

Read one ticket by id

create_ticket

Open a support ticket

update_ticket

Change a ticket's title, description, priority, due date or tags

Boards:

Tool

What it does

list_stage_options

See which stages an item is allowed to move to next

move_to_stage

Move an item to another stage on its board

Relationships, tags and comments:

Tool

What it does

link_items

Record that one item blocks, depends on or belongs under another

list_links

See what an item is linked to, and which way round

unlink_items

Remove a relationship

list_tags

List tags with their kind, so the assistant can assign the right one

add_comment

Comment on a task or ticket

list_comments

Read the comments on an item

Everything runs as you. The assistant cannot see or change anything your account cannot already see or change.

Read-only tools are marked as such, so most clients run them without asking. Tools that create or change something are marked as writes, and clients will normally ask you to confirm before they run.

Lists return at most 25 items per call, and report how many matched in total, so the assistant can ask for the next page when it needs more.

Tags are how Flow Master classifies work, and the kind tells the assistant what a tag means — for example a partner space has its own tag. Ask the assistant to tag something and it will look the tag up rather than guess. Setting tags replaces the existing ones, so ask for the full set you want.

Moving work along a board goes through the flow, not around it: the assistant asks which stages an item may move to and picks from that answer, so a move your flow does not allow simply is not offered. Say something like "this one is done, put it in review" and it will find the stage. A move can trigger the automations set up on the stage it lands in, exactly as if you had dragged the card yourself.

Relationships record how work fits together: what blocks what, what depends on what, and what sits under a parent item. Say "this one can't start until the API work is done" and the assistant records it as a dependency you can see in the app.

After a create or update, only the fields that were written come back. Ask the assistant to read the item again to see its stored state, including tags.

Commands

Command

Purpose

npx @flowmaster-io/mcp

Start the server. Your MCP client runs this for you.

npx @flowmaster-io/mcp login

Sign in through the browser

npx @flowmaster-io/mcp logout

Sign out and delete the stored session

npx @flowmaster-io/mcp whoami

Show who is currently signed in

Configuration

Variable

Required

Purpose

FLOWMASTER_API_URL

yes

Your Flow Master API URL

FLOWMASTER_WEB_URL

for login

Your Flow Master web address

FLOWMASTER_TOKEN

no

Use an existing access token instead of signing in

Neither URL has a default. If one is missing the server stops at startup with a message naming it, rather than quietly failing later.

Troubleshooting

"FLOWMASTER_API_URL is not set" — the variable is missing from your client configuration. Check the env block above.

"Not signed in" — run the login command, then restart your MCP client.

"Session expired" — sessions last 30 days. Run login again.

The assistant cannot see a project or ticket — your Flow Master account does not have access to it, or the relevant module is turned off for your organization. Ask your administrator.

A stage the assistant expected is not offered — the flow does not allow that move from the item's current stage. Move it in the app, or adjust the flow.

Sign-in works on one machine but signs you out on another — a Flow Master account holds one assistant session at a time. Signing in on a second machine ends the first.

Privacy

The server sends only what a tool call needs, and only to the Flow Master installation you configured. Your session token stays on your machine. Nothing is sent anywhere else.

Content the assistant reads from Flow Master, such as comments, is passed to whichever AI provider your MCP client uses. Your client's own privacy terms cover that step.

Support

Contact your Flow Master administrator, or open an issue at https://github.com/flow-master-io/mcp/issues

Maintenance

ActivityMaintained
ResponsivenessNo issues

Related MCP Connectors

Related MCP Servers

  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI assistants to interact with your Productboard workspace, allowing features, products, notes, objectives, key results, and releases management.
    20
    MIT
  • A
    license
    Not graded
    quality
    D
    maintenance
    Enables AI assistants to access toflow.ai workspace for prospecting, multi-channel outreach (email, LinkedIn, WhatsApp), sequences, enrichment, CRM, and AI automations.
    6
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI agents to access and manage a Workel workspace through MCP tools, reading projects, tasks, comments, events, and members, and optionally creating or updating tasks, comments, and events when write scopes and local consent are enabled.
    9 npm
    MIT