Setting Up Your First AI Scrum Master in 5 Minutes

Getting started with an AI-powered scrum master does not require a lengthy procurement process, a week of configuration, or a training seminar. ScrumChum is a GitHub Marketplace App that installs in under a minute and starts working immediately. This guide walks through the entire setup process, from installation to your first AI-powered triage, so you can see real value in your repository within five minutes.

Step 1: Install from GitHub Marketplace

ScrumChum is listed on the GitHub Marketplace. Installation follows the standard GitHub App flow:

Navigate to github.com/marketplace/scrumchum.
Click Install it for free (the Free plan includes one repository with all core features).
Choose whether to install on all repositories in your account or organization, or select specific repositories.
Review the permissions and click Install.

That is it for installation. There is no API key to generate, no webhook URL to configure, and no OAuth flow to complete. GitHub handles the authentication and permissions model, and ScrumChum receives webhook events for the repositories you selected.

A Note on Permissions

ScrumChum requests the minimum permissions needed to function: read access to issues, pull requests, and repository metadata, plus write access to issue comments (so it can post analysis results) and labels (so it can auto-label triaged issues). It does not request access to your code contents, organization members, or any administrative settings. You can review the exact permission set on the Marketplace listing before installing.

Step 2: What Happens on First Install

The moment ScrumChum is installed on a repository, several things happen automatically. Understanding this sequence helps you know what to expect and confirms that the installation is working correctly.

Configuration detection. ScrumChum checks the repository root for a .scrumchum.yml configuration file. If one does not exist, ScrumChum uses sensible defaults. You do not need a configuration file to get started; you can add one later to customize behavior.

Webhook registration. GitHub automatically delivers webhook events to ScrumChum for the installed repositories. When issues are opened, edited, or commented on, and when pull requests are created or updated, ScrumChum receives these events and can respond.

Auto-triage activation. If auto-triage is enabled (it is on by default), ScrumChum will analyze new issues as they are created and apply appropriate labels based on the issue content. The first time you open a new issue after installation, you should see ScrumChum respond within a few seconds with labels and an optional quality score.

Step 3: See Your First Auto-Triage

To verify that ScrumChum is working, create a test issue in your repository. Write a realistic issue description, something like:

Title: API rate limiting returns 500 instead of 429

When a user exceeds the API rate limit, the server returns
a 500 Internal Server Error instead of a 429 Too Many Requests
response. This causes client-side retry logic to fail because
it only handles 429 responses.

Expected: 429 status code with Retry-After header
Actual: 500 status code with generic error body
Affected endpoint: POST /api/v2/events

Within seconds of creating this issue, ScrumChum will analyze the content and post a comment. Depending on your configuration, this comment may include:

Suggested labels: Based on the issue content, ScrumChum will suggest labels like bug, api, or priority:high and apply them automatically.
Quality score: An assessment of the issue's clarity, completeness, and actionability. The test issue above is well-structured and should score well.
Triage summary: A brief AI-generated summary of the issue's nature and suggested priority.

If you see this comment, ScrumChum is installed and working correctly. If nothing appears after thirty seconds, check the repository's Settings > GitHub Apps section to confirm ScrumChum is listed and has the correct permissions.

Step 4: Try the Slash Commands

Auto-triage runs automatically, but ScrumChum's full capability set is accessed through slash commands posted as issue or PR comments. Here is a quick tour of the most useful commands to try first.

Estimate an Issue

On any open issue, post a comment with:

/scrumchum estimate

ScrumChum analyzes the issue description, acceptance criteria, and any discussion in the comments, then posts a suggested story point estimate with a brief explanation of the complexity factors it identified. This is particularly useful during sprint planning for quickly generating a starting-point estimate that the team can discuss and adjust.

Summarize a Long Thread

For issues with extensive discussion, post:

/scrumchum summarize

This generates a concise summary of the issue and all its comments, extracting the key decisions, open questions, and current status. It is invaluable for catching up on an issue you have not looked at in a few days, or for generating a quick briefing before a sprint review.

Enhance a Ticket

If you have an issue that feels underspecified, post:

/scrumchum enhance

ScrumChum reads the existing content and suggests improvements: missing acceptance criteria, additional technical context that would help the implementer, or clarifying questions that should be resolved before the issue is assigned. The suggestions appear as a comment that the author can incorporate into the issue body.

Apply Labels Intelligently

To have ScrumChum analyze an issue and apply appropriate labels, post:

/scrumchum label

This is useful for issues that were created before ScrumChum was installed, or for re-triaging an issue whose scope or nature has changed during discussion.

The Full Command Set

For a complete list of available commands, post /scrumchum help on any issue. The full set includes:

/scrumchum estimate — AI story point estimation
/scrumchum summarize — Thread summarization
/scrumchum label — Intelligent labeling
/scrumchum enhance — Ticket quality improvement suggestions
/scrumchum groom — Backlog grooming analysis (Pro)
/scrumchum metrics — Sprint and velocity metrics (Pro)
/scrumchum triage — On-demand triage analysis
/scrumchum sprint — Sprint management
/scrumchum progress — Sprint progress check
/scrumchum help — Show available commands

Commands marked "Pro" are available on the paid plan. All other commands are available on the Free plan.

Step 5: Configure with .scrumchum.yml

ScrumChum works out of the box with sensible defaults, but you can customize its behavior by adding a .scrumchum.yml file to your repository root. This file is version-controlled alongside your code, meaning configuration changes go through the normal PR review process.

Here is a starter configuration that covers the most common settings:

# .scrumchum.yml
# ScrumChum configuration — customize per repository

# Automatic triage on new issues
auto_triage: true

# Automatic label application
auto_label: true

# Quality scoring on new issues
quality_scoring: true

# Sprint configuration
sprint_length: 14          # days
standup_time: "09:00"      # 24h format
timezone: "America/New_York"

# Blocker detection thresholds
blocker_detection:
  stale_issue_days: 3
  pr_review_hours: 48
  ci_failure_hours: 24

Commit this file to your main branch, and ScrumChum will pick up the configuration immediately. Any changes to the file take effect on the next webhook event, with no restart or redeployment needed.

Key Configuration Options

auto_triage controls whether ScrumChum automatically analyzes and labels new issues. Set this to false if you prefer to invoke triage manually with the /scrumchum triage command.

quality_scoring controls whether ScrumChum posts quality scores on new issues. Some teams find this invaluable for improving ticket quality over time; others prefer to use it only on demand via /scrumchum enhance.

sprint_length and standup_time configure the sprint cadence and daily standup schedule. ScrumChum uses these to generate daily standups at the right time and to manage sprint boundaries.

blocker_detection thresholds control how aggressively ScrumChum flags potential blockers. Start with generous thresholds and tighten them as you observe your team's natural cadence.

Step 6: Upgrading to Pro

The Free plan includes one repository with full access to core features: auto-triage, quality scoring, estimation, summarization, labeling, sprint management, and blocker detection. For teams that want to use ScrumChum across multiple repositories or access advanced features, the Pro plan is available at $19 per month plus $4 per additional repository.

Pro adds the following capabilities:

Unlimited repositories: Install ScrumChum on every repo in your organization.
Retrospectives: AI-generated sprint retrospective summaries with actionable insights.
Velocity tracking: Cross-sprint velocity charts and trend analysis.
Backlog grooming: The /scrumchum groom command for automated backlog analysis and prioritization suggestions.
Sprint metrics: The /scrumchum metrics command for detailed sprint analytics.

To upgrade, visit the GitHub Marketplace listing and select the Pro plan. Billing is handled entirely through GitHub, so there is no separate account to create or credit card to enter beyond what you already have on file with GitHub.

What to Expect in Your First Sprint

After setup, here is what a typical first sprint with ScrumChum looks like:

Day 1 (Sprint Planning): Create issues for the sprint and assign them to a milestone. As each issue is created, ScrumChum auto-triages and labels it. Use /scrumchum estimate on each issue during planning to generate starting-point estimates. Use /scrumchum enhance on any tickets that feel underspecified.

Daily: ScrumChum posts a standup summary as a GitHub Issue each morning at the configured time. The standup includes yesterday's progress, today's plan based on issue assignments, and any detected blockers. Team members review the standup asynchronously and add comments if needed.

Mid-Sprint: Use /scrumchum progress to check sprint progress against the plan. ScrumChum flags any issues that are at risk of not completing on time based on current velocity and remaining scope.

Sprint End: ScrumChum generates a sprint summary with completion rate, velocity, and a list of carried-over items. On the Pro plan, it also generates a retrospective with observations about what went well, what could improve, and specific process recommendations.

The goal of the first sprint is not perfection. It is to establish the feedback loop and start collecting data. By the second or third sprint, you will have enough velocity data and quality score trends to start making meaningful process improvements backed by real numbers rather than guesses.

Five minutes to install. One sprint to see the value. That is the setup process for your AI scrum master.