About ScrumChum

Why we built this

Engineering teams deserve better than manual process overhead. We watched teams spend hours every week on tasks that should be automated: triaging incoming issues, writing standup reports, running retrospectives, and tracking sprint velocity. The data to do all of this already lives in GitHub. It just needed someone (or something) to act on it.

ScrumChum was born from a simple idea: if your code lives in GitHub, your scrum process should too. No context-switching to a separate project management tool. No copy-pasting between systems. No meetings that could have been a comment.

What ScrumChum does

ScrumChum is a GitHub App that acts as an AI-powered Scrum Master for your repositories. When you install it from the GitHub Marketplace, it observes your repository activity and automates the routine parts of agile project management:

• Issue Triage — Every new issue is automatically labeled, estimated with story points, and prioritized based on AI analysis of the title, description, and historical patterns in your repo.
• Daily Standups — Async standup summaries are generated from actual repository activity (commits, pull requests, issues) and posted on a configurable schedule.
• Quality Coaching — Every ticket is scored for clarity, acceptance criteria, and actionability. Low-scoring tickets get inline suggestions for improvement.
• Sprint Retrospectives — When a milestone closes, ScrumChum generates a structured retrospective covering velocity, highlights, blockers, and action items.
• Blocker Detection — Stale issues and stuck pull requests are surfaced before they derail your sprint, with actionable context on why they're blocked.
• Slash Commands — Nine commands (/scrumchum estimate, /summarize, /enhance, and more) let you invoke AI capabilities directly from issue comments.

Our technology

ScrumChum is built on a focused, modern stack designed for reliability and speed:

• Probot — The GitHub App framework that handles webhook delivery, authentication, and API access.
• Anthropic Claude — All AI analysis is powered by Claude, Anthropic's large language model. Claude excels at understanding technical context, structured reasoning, and producing actionable output. Anthropic does not use your data for model training.
• TypeScript — The entire codebase is TypeScript for type safety and maintainability.
• Fly.io — Application hosting with low-latency webhook processing.
• Upstash Redis — Serverless Redis for lightweight state management (installation records, usage tracking, deduplication). No issue content is stored beyond a single request.
• Cloudflare — DNS, CDN, and landing page hosting via Cloudflare Pages.

Our approach to data

We take a minimal-data approach. ScrumChum processes issue and PR metadata through Anthropic's Claude API to generate analysis, but we do not store issue content, comments, or AI responses beyond the time needed to process a single request. Installation records and usage counters are stored in Redis with automatic expiration. When you uninstall, your data is deleted.

We never access your source code. We never push commits or modify branches. The only write permission we use is posting comments and applying labels on issues. Full details are in our privacy policy.

Pricing philosophy

ScrumChum has a genuinely useful free tier: one repository with unlimited AI calls, issue triage, daily standups, quality coaching, and core slash commands. No credit card required, no time limit. We want you to see the value before you pay for anything.

The Pro plan ($19/month + $4/repo) unlocks sprint retrospectives, blocker detection, velocity tracking, and advanced commands. Billing is handled entirely through the GitHub Marketplace.

Get in touch

We'd love to hear from you:

• Support — [email protected]
• Privacy questions — [email protected]
• GitHub — ScrumChum on GitHub Marketplace