Skip to content
Back to skills

2656 Backlog Item Groomed Schema Cc5c1be3

ASecurity

**Purpose**: Define the structure of groomed content written into backlog item files. Grooming (backlog refinement) transforms items from vague ("this problem happens") to ready for planning — problem is clear, facts are verified, resources are mapped, effort is estimated, and blockers are surfaced. The agent does this autonomously: fact-checking claims, searching the codebase for related work, and identifying gaps. It does NOT produce architecture, task decomposition, or implementation plans...

  • 9 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 11, 2026
ai-agentstypescriptpythonfastapiapi

Works with

  • cli
  • api

Security analysis

A100/100

Scanned October 11, 2026

npx -y skills add tools-only/X-Skills --skill 2656-backlog-item-groomed-schema_cc5c1be3 --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of 2656 Backlog Item Groomed Schema Cc5c1be3?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for 2656 Backlog Item Groomed Schema Cc5c1be3
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tools-only-2656-backlog-item-groomed-schema-cc5c1be3/badge)](https://www.skillsdirectory.com/skills/tools-only-2656-backlog-item-groomed-schema-cc5c1be3)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

SKILL.md
# Backlog Item Groomed Schema

**Purpose**: Define the structure of groomed content written into backlog item files. Grooming (backlog refinement) transforms items from vague ("this problem happens") to ready for planning — problem is clear, facts are verified, resources are mapped, effort is estimated, and blockers are surfaced. The agent does this autonomously: fact-checking claims, searching the codebase for related work, and identifying gaps. It does NOT produce architecture, task decomposition, or implementation plans — those happen in the SAM planning phase.

**Location**: Groomed content lives in the **body** of `.claude/backlog/{priority}-{slug}.md`. Frontmatter uses the research-style `metadata:` block (aligned with `./research/` entries). Body has no duplication of frontmatter — only extra fields when present, plus `## Groomed` when groomed.

---

## Frontmatter (Research-Style)

Aligned with [research/README.md](../../research/README.md) metadata format:

```yaml
---
name: "Item title"
description: "Main problem statement or description"
metadata:
  topic: "slug-from-title"
  source: "Source description"
  added: "YYYY-MM-DD"
  priority: P0|P1|P2|Ideas
  type: Feature|Bug|Refactor|Docs|Chore
  status: open|in-progress|done|resolved
  # optional:
  issue: "#N"
  plan: "path/to/plan.md"
  groomed: "YYYY-MM-DD"
  layer: "0"|"1"|"2"
  language: python|typescript|...
  stack: fastapi|python-cli|...
---
```

---

## Body (No Duplication)

Body is **empty** when un-groomed and no extra details. When present, body contains only:

- **Suggested location** — when applicable
- **Research first** — when applicable
- **Decision needed** — when applicable
- **Files** — when applicable
- **## Groomed (YYYY-MM-DD)** — when groomed (see below)

Do **not** duplicate `name`, `description`, `source`, `added`, `priority`, `type`, `issue` in the body — they live in frontmatter.

---

## Groomed Sections (Body, under ## Groomed)

| Section | Purpose | Required |
|---------|---------|----------|
| **Reproducibility** | Steps x, y, z to replicate the problem | When applicable |
| **Output / Evidence** | Steps to see the issue; screenshot or log references | When applicable |
| **Priority** | Numeric (e.g., 8/10) with rationale | Yes |
| **Impact** | Bottleneck to x, y, z; who/what is blocked | When applicable |
| **Benefits** | What doing this unlocks (x, y, z) | When applicable |
| **Expected Behavior** | How it should work | Yes |
| **Desired Structure** | The target state we want | When applicable |
| **Acceptance Criteria** | Concrete checks for "done" | Yes |
| **Human Input** | Output of interviewing the human partner; desired outcome | When BLOCKED or human input needed |
| **Questions for Human** | Prompts to ask when info is missing (ARL human-probing) | When BLOCKED |
| **Resources** | Compact table: skills, agents, prior work | Yes |
| **Dependencies** | Items this depends on; items this unblocks | When applicable |
| **Blockers** | Missing prerequisites; RT-ICA BLOCKED reason | When BLOCKED |
| **Effort** | Small / Medium / High | When estimable |

---

## RT-ICA Integration

- **APPROVED**: Item is ready; groomed sections filled
- **BLOCKED**: Include **Blockers** and **Questions for Human**; do not treat as ready for planning

---

## Example Groomed Body

```markdown
## Description

Original problem statement from backlog.

## Groomed (2026-02-23)

### Reproducibility

1. Run `uv run prek run --files foo.py`
2. Observe LK001 on line 42

### Output / Evidence

- Screenshot: `.claude/backlog/assets/p1-foo-screenshot.png`
- Log excerpt in `plugins/plugin-creator/planning/qa-report.md`

### Priority

8/10 — Blocks plugin validation CI gate; affects all contributors.

### Impact

- Blocks: manifest-sync job, pre-commit hooks
- Bottleneck: Every PR runs plugin_validator

### Benefits

- CI gate becomes reliable
- Pre-commit output actionable

### Expected Behavior

Plugin validator should report unique files, not validator invocations.

### Acceptance Criteria

1. Running validator on 1 file shows "Total files: 1"
2. Summary counts unique files

### Resources

| Type | Item |
|------|------|
| Skill | plugin-creator:claude-hooks-reference-2026 |
| Agent | @refactor-executor |
| Prior work | plan/tasks-2-validator-ux-coverage.md |
```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.