SLOPE

Sprint Lifecycle & Operational Performance Engine

A framework for measuring and improving sprint execution quality. Replace subjective retrospectives with objective, quantifiable metrics. Works with Claude Code, Cursor, and OpenCode.

Quick Start

# Install globally (makes `slope` available everywhere)
npm install -g @slope-dev/slope

# Or install locally in your project and use npx
npm install @slope-dev/slope
npx slope init

# Initialize in your project (auto-detects your AI tool)
slope init

# Or specify your platform
slope init --claude-code
slope init --cursor
slope init --opencode

# View your handicap card
slope card

# Get a pre-sprint briefing
slope briefing

# Generate an HTML performance report
slope report --html

Note: If installed locally (without -g), prefix commands with npx — e.g., npx slope card.

What is SLOPE?

SLOPE maps sprint execution to a scoring metaphor. Every sprint has a par (expected ticket count), every ticket is a shot with an approach complexity and outcome. Over time, your handicap reveals patterns: Do you over-engineer? Under-scope? Pick the wrong approach?

SLOPE provides:

• Scorecards — structured sprint retros with quantified outcomes
• Handicap tracking — rolling performance windows (last 5, 10, all-time)
• Dispersion analysis — miss pattern detection and systemic issue identification
• Training recommendations — data-driven improvement suggestions
• Agent guidance hooks — real-time hints for AI coding agents
• HTML reports — self-contained visual performance dashboards
• Multi-platform support — Claude Code, Cursor, OpenCode

Metaphors

SLOPE ships with 6 built-in metaphors. The scoring math is identical — only the terminology changes.

Metaphor	Sprint	Ticket	Perfect	Par	Miss
Golf (default)	Sprint	Shot	Hole-in-One	Par	Missed Long/Short/Left/Right
Tennis	Set	Point	Ace	Deuce	Wide/Net/Long/Out
Baseball	Inning	At-Bat	Home Run	Single	Foul/Strike/Pop/Ground
Gaming	Level	Quest	S-Rank	B-Rank	Over-leveled/Under-leveled/Wrong Path/Side-tracked
D&D	Quest	Encounter	Natural 20	DC Met	Fumble/Misfire/Detour/Distraction
Matrix	Simulation	Anomaly	The One	Stable	Overclocked/Underclocked/Drift/Noise

Configure in .slope/config.json:

{ "metaphor": "gaming" }

Or per-command: slope review --metaphor=tennis

Package

Published as a single package: @slope-dev/slope

Includes the core scoring engine, SQLite store, CLI (30 commands), and MCP server.

CLI Commands

Scoring & Analysis

Command	Description
slope card	Display handicap card with rolling windows
slope validate [path]	Validate scorecard(s)
slope review [path] [--plain] [--metaphor=id]	Format sprint review as markdown
slope review recommend	Recommend review types for current sprint
slope review findings add|list|clear	Track implementation review findings
slope review amend [--sprint=N]	Apply review findings as hazards and recalculate score
slope report --html [--output=path] [--metaphor=id]	Generate HTML performance report
slope tournament --id=<id> --sprints=N..M	Build tournament review from sprint range
slope auto-card --sprint=N [--ci=path]	Generate scorecard from git + CI signals

Planning & Guidance

Command	Description
slope briefing [--sprint=N] [--categories=...] [--keywords=...]	Pre-sprint briefing
slope plan --complexity=<level>	Club recommendation + training plan
slope classify --scope=... --modified=... --tests=pass	Classify a shot from execution trace
slope next	Show next sprint number
slope roadmap validate|review|status|show	Strategic planning tools

Sessions & Claims

Command	Description
slope session start|end|heartbeat|list	Manage live sessions
slope claim --target=<t> [--scope=area] [--force]	Claim a ticket or area
slope release --id=<id>	Release a claim
slope status [--sprint=N]	Show sprint course status + conflicts

Agent Hooks

Command	Description
slope hook add|remove|list|show	Manage lifecycle hooks
slope hook add --level=full	Install all guidance hooks
slope guard list|enable|disable	Manage guard activation
slope guard <name>	Run a guard handler (stdin/stdout)

Setup

Command	Description
slope init	Create .slope/ directory with config
slope init --claude-code	Install Claude Code rules, hooks, MCP config
slope init --cursor	Install Cursor rules + MCP config
slope init --opencode	Install OpenCode AGENTS.md + plugin
slope init --all	Install for all detected platforms
slope extract --file=<path>	Extract events into SLOPE store
slope distill [--auto]	Promote event patterns to common issues

Platform Setup

Claude Code

slope init --claude-code

Installs:

• .claude/rules/ — Sprint checklist, commit discipline, review loop
• .claude/hooks/ — Guard dispatcher for real-time guidance
• .mcp.json — SLOPE MCP server config
• CLAUDE.md — Project context

Cursor

slope init --cursor

Installs:

• .cursor/rules/ — SLOPE methodology rules
• .cursor/mcp.json — MCP server config

OpenCode

slope init --opencode

Installs:

• AGENTS.md — SLOPE methodology (OpenCode reads this format)
• .opencode/plugins/slope-plugin.ts — Event capture plugin

Agent Guidance Hooks

SLOPE can guide AI agents in real-time via hook integration. 15 guards provide contextual hints:

Guard	Trigger	What it does
explore	Before search/read	Suggests checking codebase index first
hazard	Before file edit	Warns about known issues in the area
commit-nudge	After file edit	Nudges commit/push after prolonged editing
scope-drift	Before file edit	Warns when editing outside claimed scope
compaction	Before context compact	Saves checkpoint to store
stop-check	Before session end	Blocks if uncommitted/unpushed work exists
next-action	Before session end	Suggests next actions based on sprint state
subagent-gate	Before task spawn	Enforces model and turn limits on subagents
push-nudge	After tool use	Nudges push after prolonged period without pushing
workflow-gate	Before tool use	Enforces sprint workflow compliance
review-tier	Before tool use	Suggests review tier based on sprint complexity
version-check	Before tool use	Warns when SLOPE version is outdated
stale-flows	Before file edit	Warns when editing files in stale flow definitions
pr-review	After tool use	Suggests PR review actions
transcript	After tool use	Captures session transcript data

Install all guards:

slope hook add --level=full

Guards are non-blocking hints (except stop-check) — they inject context, never deny actions.

Configuration

After slope init, configure .slope/config.json:

{
  "scorecardDir": "docs/retros",
  "scorecardPattern": "sprint-*.json",
  "minSprint": 1,
  "metaphor": "golf",
  "commonIssuesPath": ".slope/common-issues.json",
  "sessionsPath": ".slope/sessions.json",
  "roadmapPath": "docs/backlog/roadmap.json",
  "guidance": {
    "disabled": [],
    "commitInterval": 15,
    "pushInterval": 30
  }
}

Core API

import {
  buildScorecard,
  validateScorecard,
  computeHandicapCard,
  computeDispersion,
  formatSprintReview,
  recommendClub,
  classifyShot,
  generateTrainingPlan,
  buildReportData,
  generateHtmlReport,
  getMetaphor,
  loadScorecards,
  recommendReviews,
  findingToHazard,
  amendScorecardWithFindings,
} from '@slope-dev/slope';

// Build a scorecard
const card = buildScorecard({
  sprint_number: 1,
  theme: 'My First Sprint',
  par: 3,
  slope: 0,
  date: '2026-02-22',
  shots: [
    { ticket_key: 'S1-1', title: 'Setup', club: 'short_iron', result: 'green', hazards: [] },
    { ticket_key: 'S1-2', title: 'Feature', club: 'short_iron', result: 'in_the_hole', hazards: [] },
    { ticket_key: 'S1-3', title: 'Tests', club: 'wedge', result: 'green', hazards: [] },
  ],
});

// Generate HTML report with gaming metaphor
const data = buildReportData([card]);
const gaming = getMetaphor('gaming');
const html = generateHtmlReport(data, gaming);

License

MIT