Kuriflow Mcp

io.github.sGuNk13Published by GitHub account sGuNk13v0.1.1Updated Oct 24, 2026

Make AI tasks repeatable. Run with new data automatically — zero AI tokens per run.

VerifiedSTDIODesktop onlyProductivity & WorkflowAI & ML

Overview

AI-generated overview

Kuriflow MCP lets an assistant save a one-off AI task as a repeatable workflow that later runs on new data without spending AI tokens per run.

What you can do
Save the assistant's solution as a repeatable workflow, and list saved workflows and available workflow types. Run a workflow again with new data, then check execution results and download outputs. Set runs to trigger from email attachments, a Google Drive folder, a schedule or a manual upload. Request human approval before proceeding and check the approval decision. Look up regulatory rules.
When to use it
Use it when a task your assistant can already solve has to be repeated with fresh data, such as recurring reports, spreadsheet analysis or template filling. For a one-off analysis, asking the assistant directly is enough.
Requirements
Runs locally via uvx from the PyPI package kuriflow-mcp, so a Python environment is needed; a remote endpoint with bearer auth is also offered. Needs a Kuriflow account (Google sign-in) and KURIFLOW_API_KEY; KURIFLOW_API_URL defaults to the hosted backend and only needs overriding for self-hosting. Network access required.
Before you install
Requires a Kuriflow API key (KURIFLOW_API_KEY), shown only once, so store it safely. Workflows read from your Gmail or Google Drive and send that data to the Kuriflow backend for processing. Saving and running workflows creates and changes data in your account. The free tier allows 10 runs per month; more requires contacting support.

Installation

In SourceWeft

  1. Open Kuriflow Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

mcp-name: io.github.sGuNk13/kuriflow-mcp

Kuriflow MCP Server

AI solves tasks once. Kuriflow makes them repeat.

Stop spending AI tokens on the same task every week. Let your AI agent solve it once, then Kuriflow runs it automatically with new data — via email, Google Drive, or schedule. Zero token cost per run.

Quick Start

Option 1: Claude Cowork — one-click OAuth (recommended)

If you use Claude Cowork (Pro or Max plan):

  1. Settings → Connectors → Add custom connector
  2. Paste the server URL: https://mcp.kuriflow.com
  3. Click Connect
  4. Sign in with Google (Kuriflow uses your Google account — no separate signup)
  5. Click Allow on the consent screen

Tools become available in any new chat. Zero copy-paste, zero JSON.

Option 2: Remote with API key

For Claude Desktop or any client that supports raw remote MCP servers with bearer auth. Edit your client's MCP config:

json
{  "mcpServers": {    "kuriflow": {      "url": "https://mcp.kuriflow.com",      "headers": {        "Authorization": "Bearer YOUR_API_KEY"      }    }  }}

Option 3: Local install (uvx)

json
{  "mcpServers": {    "kuriflow": {      "command": "uvx",      "args": ["kuriflow-mcp"],      "env": {        "KURIFLOW_API_KEY": "your_kf_key",        "KURIFLOW_API_URL": "https://api.kuriflow.com"      }    }  }}

Get Your API Key (Options 2 & 3 only)

Option 1 (Cowork OAuth) does NOT require an API key — skip this section.

  1. Sign up at kuriflow.com/signup (30 seconds, Google account)
  2. Open Settings → API Keys in your dashboard, click Generate, and copy the shown key — it's displayed only once
  3. Add it to your config above

How It Works

  1. Ask your AI agent (Claude, Gemini, ChatGPT, or any MCP-compatible client) to solve a task. For example:
    • "Analyze this sales data and create a dashboard with charts"
    • "Build a DCF model from the financial statement fetched from my Google Drive"
    • "Generate a monthly expense report from this spreadsheet using our template"
  2. The AI gives the solution.
  3. You ask the AI to save it as a repeatable workflow with Kuriflow.
  4. Choose how new data arrives:
    • Email trigger — Kuriflow watches your Gmail. When someone sends an attachment, the workflow runs automatically.
    • Google Drive — Kuriflow checks a Drive folder on your schedule (hourly, daily, weekly). New file? Workflow runs.
    • Schedule only — No input needed. The workflow fetches its own data from online sources and runs on your schedule.
    • Manual upload — Upload a file whenever you're ready.

You can also provide template files (.docx, .xlsx, .pptx) so Kuriflow fills them with fresh data every run.

Tools

ToolDescription
save_workflowSave the AI's solution as a repeatable workflow
run_workflowRun it again with new data
list_workflowsSee your saved workflows
list_kurisBrowse available workflow types
get_execution_resultCheck results and download outputs
request_approvalRequest human approval before proceeding
get_approval_statusCheck approval decision
query_regulationLook up regulatory rules

Why Kuriflow?

AI aloneAI + Kuriflow
First timeAI solves itAI solves it
Every repeatTokens burned againRuns with new data, no tokens
Monthly costGrows with every repeatFixed after setup
Your tokensSpent on routineSaved for what matters

Free Early Access

10 runs per month. Need more? Email [email protected].

Links

License

MIT

Source: README.md at commit b06fd19

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.1LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts