ScrumDo
See plans
ScrumDo RunnerDeveloper Preview

Run governed agents on your own machine.

ScrumDo assigns the work; the runner executes it on your computer, from the card. Keep the specification, plan, execution, QA, and proof attached to the record — on your compute, with your keys.

Checking the latest release…
Download for macOSDownload for WindowsLinux & CLI setup
  • Claude
  • Codex
  • Grok
  • Bring your own keys
Managed execution from the card

ScrumDo starts the run; your machine does the work.

The runner is a supervised process on your computer that picks up work from ScrumDo and executes it through the governed lifecycle — nothing runs that the card did not assign.

01

Runs start from the card

ScrumDo assigns work to an agent and manages the specification, plan, execution, QA, proof, and human acceptance from the work record — not from a private chat.

02

Your machine, your compute

The runner works in your local checkouts and only makes outbound calls to ScrumDo. Your connect credential and any bring-your-own keys stay in the OS keychain.

03

The lifecycle stays visible

Every step — plan, execution, verification, proof — is attributable in ScrumDo, and consequential approvals remain explicit human decisions.

Two ways to bring agents to the work

Runner and MCP solve different interaction problems.

Work starts from the ScrumDo card

Use ScrumDo Runner

You want ScrumDo to assign work to an agent and manage the specification, plan, execution, QA, proof, and human acceptance from the card.

  • Best for managed card-level runs
  • ScrumDo starts and tracks execution
  • Runs on your machine and your compute
Work starts from your AI tool

Use ScrumDo MCP

You are already working in Codex, Claude Code, or Cursor and want that assistant to read and act on permitted ScrumDo work directly.

  • Best for developer-led interactive work
  • Your selected client starts the request
  • ScrumDo remains the shared work record

Use both when developers want an interactive assistant while ScrumDo manages the governed run lifecycle and keeps the shared record. See ScrumDo MCP →

Get ScrumDo Runner

The desktop app on macOS, or the CLI on Linux and servers.

macOS

macOS 12 Monterey or later · Apple Silicon (arm64)
Download .dmg

Signed and notarized by Apple — installs with no Gatekeeper warning.

Windows

Windows 10 / 11 · 64-bit
Download installer

Setup .exe (an .msi is also on the release). Unsigned for now, so Windows SmartScreen may warn — choose More info → Run anyway.

Linux & servers

Python 3.11+ · the spryng-agent CLI

Same runner, headless — ideal for Linux boxes, servers and CI.

macOS: all releases · Linux/CLI: spryng-agent v0.1.22 on PyPI

Complete Linux / CLI setup guideprerequisites → install → connect → run · about 5 minutes

  1. Check prerequisites

    You need Python 3.11+, git, and pipx. Most distros already ship Python and git.

    python3 --version && git --version

  2. Install pipx

    pipx installs the CLI into an isolated environment and puts it on your PATH.

    python3 -m pip install --user pipx && python3 -m pipx ensurepath

    On Debian/Ubuntu you can instead run: sudo apt install pipx. Open a new shell afterwards.

  3. Install the runner

    Install the spryng-agent CLI from PyPI.

    pipx install spryng-agent

  4. Sign in to your engine

    The runner drives an engine CLI you already use — Claude Code, Codex or Grok. Install one and sign in so it is authenticated and on your PATH.

    Grok is bring-your-own-key: export XAI_API_KEY=<your xAI key>. Claude and Codex use their own login.

  5. Connect to ScrumDo

    Open your runtime in ScrumDo, copy its Runtime ID and connect credential, then run login and paste them when prompted.

    spryng-agent login

    Stored in ~/.spryng/agent.toml with 0600 permissions — never a world-readable file.

  6. Verify the setup

    doctor checks the engine CLI, git, engine authentication, and that ScrumDo is reachable before any card runs.

    spryng-agent doctor

  7. Start the runner

    Start the pull loop — it picks up work from ScrumDo and runs it on your machine. Single repo: cd in and add --use-cwd; multiple: map each with spryng-agent workspace add owner/name /path.

    spryng-agent run

Your connection is saved to ~/.spryng/agent.toml. If a step fails, spryng-agent doctor tells you exactly what to fix.

Before you run

Where the work runs, and what stays on your machine.

What is the difference between the app and the CLI?

They run the same runner. The desktop app is a menu-bar UI that supervises one or more runtimes with configuration, status, and readiness checks. The spryng-agent CLI is the same runner for the terminal — ideal for Linux, servers, and CI.

Does my code leave my machine?

No. The runner works in your local checkouts and only makes outbound calls to ScrumDo to pick up work and report status. There are no inbound connections.

Where are my credentials stored?

The connect credential and any bring-your-own API keys are kept in your operating system keychain — never in a plaintext config file. The runner is outbound-only.

Which engines are supported?

Claude, Codex, and Grok. You point each runtime at the engine CLI you already have installed and, if you use your own keys, add them per runtime.

How is ScrumDo Runner different from ScrumDo MCP?

Use Runner when ScrumDo should start and manage an agent run from the card — specification, plan, execution, QA, proof, and human acceptance. Use MCP when you are already working in an AI tool and want that assistant to interact with ScrumDo directly.

Is the download safe to install?

Yes. The macOS app is signed with an Apple Developer ID and notarized by Apple, so it installs without a Gatekeeper warning. On Linux, Windows, and servers you can run the same runner today with the spryng-agent CLI.

Your machine, one accountable record

Bring your runtimes online.

Download the app or install the CLI, connect a runtime, and let ScrumDo drive governed runs on your own compute.

Download for macOSSee ScrumDo MCP