Open Source Sprint Scoring

AI agents don't learn from their mistakes.

AI agents can write code, run tests, and ship features. But they start every session from scratch — no memory of what worked, what failed, or what to do differently.

SLOPE gives them a framework that produces data humans can actually use.

For anyone using AI to build software — and wondering why it doesn't seem to get better.

27 sprints and counting
See how it works

AI Agents Are Powerful. But Chaotic.

Without structure, session 50 looks exactly like session 1. Work gets lost. Patterns repeat. Nobody knows if things are getting better.

AI agents build real software.

They write code, run tests, fix bugs, and ship features. They're fast, capable, and increasingly common.

They work in sessions.

You describe what you want. The agent executes. It delivers. Session over.

But nothing carries over.

Each session starts from zero. No memory of past mistakes. No awareness of patterns. No measurement of progress.

After 50 sessions, you know exactly as much as you did after the first.

All that work — and no data to show what's improving, what's breaking, or what to do next.

Without SLOPE

ERROR: Agent crashed mid-sprint, no recovery point

$ git log --oneline | wc -l → 47 commits, no pattern

WARN: Same bug as 3 sprints ago

$ how many tests? → "not sure, maybe 200?"

ERROR: Context lost on compaction

$ what improved? → "hard to say"

... (scrolls for 500 more lines)

With SLOPE

Sprint 27 — Expected: 4 tickets — Delivered: 4 (on target)

Clean starts: 4/4 | Completed as planned: 3/4

Known issues hit: 1 — caught by automated guard

Trending: improving (last 5 better than all-time)

Recurring pattern flagged → won't happen again

Next sprint: guided by briefing + performance history

27
Sprints
1,091
Tests
26
CLI Commands
5
Packages

How SLOPE Works

Structure. Methodology. Language.

SLOPE layers a framework on top of agent work so the output is useful to humans — not just "done" or "not done", but structured data you can learn from.

1

Structure

SLOPE organizes agent work into sprints — focused batches with a clear scope, a beginning, and an end. Instead of an endless stream of sessions, each batch becomes a unit you can measure.

2

Methodology

Each sprint follows a repeatable routine: set expectations, declare difficulty, execute, score the result, review what happened. This turns raw work into structured data — how hard it was, what went wrong, and whether patterns are emerging.

3

Language

Then SLOPE wraps that data in a metaphor your team already understands. Golf, gaming, D&D — the metaphor makes the data intuitive. Same engine, different vocabulary.

The scoring engine doesn't change. Only the language does.

Raw Work

Sessions

What agents produce

SLOPE

Hole · Level · Encounter

How it gets framed

You Gain

Insight

Data you can act on

See It In Action

Pick the language your team already speaks

Same sprint. Same engine. Same data. Watch the vocabulary change.

Sprint 27 Summary

Par 4, Score 4

Fairway: 4/4 | GIR: 3/4 | Bunkers: 1

Handicap: 1.2 — trending down

Same underlying data. Different vocabulary. The engine is identical — the language is a skin.

It Works

Consistent routines compound.

Performance Trajectory
S5
3.2
S10
2.5
S20
1.8
S27
1.2
Lower is better — rolling performance average over 27 sprints

In Sprint 12, the same bug appeared for the third time. SLOPE flagged it as a recurring hazard. It never appeared again.

Every metric on this page comes from the reference implementation — 27 sprints of real work.

Quick Start

Four commands. First sprint.

From zero to your first scored sprint. No configuration required — SLOPE ships with sensible defaults.

$ npm install -g @srbryers/cli
1
slope init

Initialize

Set up SLOPE in your project. Pick a metaphor. Configure guards.

2
slope briefing

Get briefed

Get a pre-sprint briefing: risk index, performance snapshot, known issues.

3
slope card

Check your card

Check your performance card — rolling stats across multiple windows.

4
slope review

Review the sprint

Generate a sprint review with ticket-by-ticket analysis and improvement recommendations.

Go Deeper

Ready to score your sprints? Start with the CLI. Want the full technical breakdown? Read the framework.

Start scoring your sprints

$ npm install -g @srbryers/cli

Then run slope init in your project

Read the full framework

Scoring engine, metaphor system, guards, routines, and the complete methodology.

The SLOPE Framework

@srbryers/core · @srbryers/cli · @srbryers/tokens · @srbryers/mcp-tools

SLOPE is an open-source sprint scoring framework for AI agent teams.

Built with Astro, Tailwind, and GSAP. Live stats from the reference implementation.