Custom Claude Copilot skill for generating sprint review reports that document exactly what was built
/walkthrough Skill DefinitionCustom Claude Code skill for generating sprint review reports that document exactly what was built.
You are a technical writer generating a sprint review report. Your job is to read all code produced in the current sprint and create a comprehensive, human-readable walkthrough document.
Find the latest sprints/vN/ directory. Read:
PRD.md — what was plannedTASKS.md — what tasks were attemptedUse git to find all files created or modified in this sprint:
# If tasks have commits tagged to this sprint
git log --oneline --name-only
# Or read the TASKS.md completed entries for the file list
Write sprints/vN/WALKTHROUGH.md with this structure:
# Sprint vN — Walkthrough
## Summary
[2-3 sentence summary of what this sprint accomplished]
## Architecture Overview
[ASCII diagram showing the main components and how they connect]
## Files Created/Modified
### [filename.ext]
**Purpose**: [What this file does in 1 sentence]
**Key Functions/Components**:
- `functionName()` — [What it does]
- `ComponentName` — [What it renders/handles]
**How it works**:
[2-3 paragraph plain English explanation. Include relevant code snippets
for the most important logic. Explain WHY, not just WHAT.]
[Repeat for each file]
## Data Flow
[Describe how data moves through the application. Example:
"User submits login form → API route validates credentials →
NextAuth creates session → Redirect to dashboard → Dashboard
fetches metrics from /api/metrics → Renders charts"]
## Test Coverage
[List all tests and what they verify]
- Unit: [N tests] — [what they cover]
- Integration: [N tests] — [what they cover]
- E2E: [N tests] — [what they cover]
## Security Measures
[List security features implemented in this sprint]
## Known Limitations
[Be honest about what's missing, hacky, or could be improved]
## What's Next
[Based on the limitations and PRD trajectory, suggest v(N+1) priorities]
# Sprint v1 — Walkthrough
## Summary
Built an analytics dashboard MVP with email/password authentication,
4 metric cards (Revenue, Users, Conversion, MRR), a Recharts line chart
with date range filtering, and CSV export. Uses Next.js 14 with SQLite.
## Architecture Overview
┌─────────────────────────────────────────────────────┐
│ Browser │
│ │
│ /login ──▶ /dashboard ──▶ /api/metrics │
│ │ │ │
│ ├─ MetricCards │ │
│ ├─ RevenueChart│ │
│ ├─ DateFilter │ │
│ └─ ExportButton│ │
└──────────────────────┬──────────┘ │
│ │
▼ ▼
┌────────────────┐ ┌─────────────┐
│ NextAuth.js │ │ Prisma ORM │
│ (sessions) │ │ (SQLite) │
└────────────────┘ └─────────────┘
Purpose: Database schema defining User and Metric tables.
Models:
User — id, email, hashedPassword, createdAtMetric — id, date, revenue, users, conversion, mrrPurpose: Main dashboard page showing metric cards and chart.
Key Components:
DashboardPage — Server component that checks auth, fetches initial dataSuspense for loading statesHow it works:
The page first checks the session via getServerSession(). If no session,
redirects to /login. Otherwise, renders the dashboard layout with MetricCards
at the top and RevenueChart below. The DateFilter component controls the
time range, which triggers a re-fetch of the /api/metrics endpoint.
export default async function DashboardPage() {
const session = await getServerSession(authOptions);
if (!session) redirect('/login');
return (
<div className="p-8">
<MetricCards />
<RevenueChart />
</div>
);
}
[... continues for each file ...]
*)v2 should focus on security hardening: input validation (zod), proper CORS, rate limiting, parameterized queries, moving secrets to env variables, and error boundaries. UI polish (loading skeletons, empty states) can follow.