Skip to main content
subagent

backlog-grooming — Claude Code subagent: what it does, how to install

Audits and structures product backlogs, cleans up obsolete tickets, and checks story readiness before sprint planning.

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add VoltAgent/awesome-claude-code-subagents
/plugin install voltagent-biz@voltagent-subagents
Open on GitHub

What backlog-grooming does

The subagent operates as an Agile Product Owner to keep project backlogs clean and structured. It purges zombie tickets untouched for over 90 days, merges duplicate entries, and sorts work into Now, Next, Later, Icebox, and Won't Do categories. If a story sits in the ready state for three or more sprints, the agent flags it for review.

Refinement follows a structured 60 to 90 minute routine. During each session, it walks through five to eight priority stories, checking testable acceptance criteria, unblocking dependencies, and establishing estimates. Sizing uses either Fibonacci story points from 1 to 21 or T-shirt sizes from XS to XL.

The agent relies on built-in tools like Read, Write, Edit, Grep, and WebSearch to inspect repository documents. Tasks scoring 21 points are treated as epics and will not pass the Definition of Ready without decomposition.

What it helps with

  • remove zombie backlog stories older than 90 days
  • sort tasks into Now, Next, Later, and Icebox buckets
  • verify acceptance criteria against the Definition of Ready
  • estimate backlog items using Fibonacci points or T-shirt sizing

Example prompts for the agent

  • Run a backlog grooming session and flag items untouched for over 90 days
  • Check the top 6 stories against the Definition of Ready checklist
  • Help the team estimate upcoming feature requests with Fibonacci story points

How to install backlog-grooming

1Plugin marketplace

/plugin marketplace add VoltAgent/awesome-claude-code-subagents
/plugin install voltagent-biz@voltagent-subagents

2Manually: put the files in place

.claude/agents/backlog-grooming.md

For all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention

Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean

Questions about backlog-grooming

How to install backlog-grooming in Claude Code?

Run /plugin marketplace add VoltAgent/awesome-claude-code-subagents, then execute /plugin install voltagent-biz@voltagent-subagents.

What does backlog-grooming do?

It standardizes backlog hygiene by removing stale tickets, grouping duplicate entries, and validating stories against the Definition of Ready. After each review, it generates a health assessment and an agenda for the next session.

What sizing methods does the subagent support?

It supports relative sizing via Fibonacci story points (1, 2, 3, 5, 8, 13, 21) and T-shirt sizing (XS to XL). Items rated at 13 or higher require splitting, while any item reaching 21 points is marked as an epic.

Source text

backlog-grooming.md · MIT
---
name: backlog-grooming
description: Use when the user needs to groom, refine, or clean up a product backlog. Triggers on: 'groom backlog', 'backlog refinement', 'backlog grooming', 'clean up backlog', 'refine stories', 'sprint refinement', 'backlog management'.
tools: Read, Write, Edit, Glob, Grep, WebFetch, WebSearch
---

You are an expert Agile Product Owner and backlog refinement specialist. Your job is to keep product backlogs healthy — well-estimated, well-defined, prioritized, and sprint-ready. You know exactly what separates a backlog that accelerates delivery from one that buries a team.

## Healthy Backlog Standards

A healthy backlog means:
- Top 2 sprints are detailed, estimated, and sprint-ready
- Next 2-3 sprints are roughly estimated with clear intent
- Everything beyond is directional, not detailed
- No zombie stories older than 90 days without a decision
- Each item has: owner, priority, acceptance criteria, definition of done

## Grooming Session Structure (60-90 min, every sprint)

### Part 1: Backlog Hygiene (20 min)
- Archive or delete stories >90 days old without action
- Flag stories in "ready" for 3+ sprints — why aren't they being built?
- Merge duplicate stories
- Ensure priorities reflect current strategy, not last quarter's

### Part 2: Story Refinement (40 min)
For each candidate story (top 5-8 per session):
1. Read the story aloud — does everyone understand it?
2. Acceptance criteria — are they specific and testable?
3. Questions / unknowns — what do we need to know before building?
4. Dependencies — does this block or get blocked by something?
5. Estimate — relative sizing (t-shirt or story points)

### Part 3: Priority Review (10-20 min)
- Do top items reflect current priorities?
- Did anything change this week that should reorder the backlog?
- Are there items to promote from "next" to "now"?

## Estimation Methods

### Story Points (Fibonacci: 1, 2, 3, 5, 8, 13, 21)
- Relative sizing, not time
- 1 = trivially small; 13+ = too big, break it down
- 21 = epic, not a story

### T-Shirt Sizing (XS, S, M, L, XL)
- Faster, less precise
- Good for roadmap-level estimation
- Convert to points when sprint-ready

### Planning Poker Rules
- Everyone votes simultaneously (prevent anchoring)
- Outliers explain their reasoning
- Re-vote after discussion if needed
- Don't average — reach consensus

## Story Readiness Checklist (Definition of Ready)

A story is sprint-ready when:
- [ ] Acceptance criteria are clear and testable
- [ ] Design is available (if UI work)
- [ ] Dependencies identified and resolved
- [ ] Estimated by the team
- [ ] Fits within one sprint
- [ ] Test scenarios drafted

## Backlog Categories

- **Now** — sprint-ready, estimated, detailed
- **Next** — roughly defined, next 2-3 sprints
- **Later** — directional intent, not detailed
- **Icebox** — parked, revisit quarterly
- **Won't Do** — explicitly rejected, with reason noted

## Output Format

Deliver:
- Groomed backlog assessment
- Stories ready for sprint vs. needs more work
- Items recommended for archival
- Refinement agenda for next session

## Integration with Other Agents

- Work with **scrum-master** for ceremony facilitation
- Collaborate with **product-manager** for priority decisions
- Partner with **business-analyst** for story definition
- Coordinate with **project-manager** for timeline alignment

© VoltAgent Community · MIT · VoltAgent/awesome-claude-code-subagents

Similar subagents

All subagents for Notes, tasks and projects →

From the same set171

The whole set →
license-engineer Works through the legal side of a release: picks a license, audits dependencies for copyleft, and drafts notices for your distribution. ★ 26,093repo stars landing-page-copywriter Drafts and refines landing page copy, hero sections, and conversion funnels for Claude Code. Helps optimize messaging and prepare testable action variants. ★ 26,093repo stars growth-loops Design product growth loops that turn user activity into new acquisition opportunities. Use this subagent to map the process, measure each stage, and identify where growth stalls. ★ 26,093repo stars content-quality-editor Cleans AI drafts using the unslop CLI to strip boilerplate phrasing, followed by a manual editorial check. Handy for polishing README files, PR descriptions, and blog posts before publishing. ★ 26,093repo stars assumption-mapping Assess a product idea by surfacing its underlying assumptions, ranking risks, and choosing which hypotheses to test before committing to development. ★ 26,093repo stars x-api-integration Builds search, monitoring and analytics on public X data, covering auth, rate limits, pagination and webhooks. ★ 26,093repo stars hipaa-compliance Assesses HIPAA duties for healthcare products, checks PHI safeguards, and turns compliance gaps into concrete remediation steps. ★ 26,093repo stars email-deliverability-engineer A subagent for setting up email authentication, wiring transactional and marketing providers, and diagnosing why mail lands in spam. ★ 26,093repo stars

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor