Skip to content
Beta Multi-model support (OpenAI, Gemini, and more) launching soon.
RBA RBA Skills

perf-backlog-builder

Installable Ships in performance-auditor — or copy the raw skill below.
Open Claude.ai

Paste into Claude Code, Claude.ai, the desktop app, or the browser extension — ⌘/Ctrl+V.

ExperimentalNo security review

Performance Backlog Builder

Convert a performance audit report into a sprint-ready backlog with severity-sequenced delivery. Works with output from any perf-* audit skill in this plugin.

Usage

/perf-backlog-builder <audit-report-path> [--format ado|jira|markdown] [--sprints <n>] [--team <team-name>]

Examples:

  • /perf-backlog-builder ./cwv-audit-report.md --format ado --sprints 2
  • /perf-backlog-builder ./full-audit.md --format jira --team "Digital Engineering"
  • /perf-backlog-builder ./findings.md --format markdown — Markdown table for review

Phase 1 — Ingest and Triage

Read the audit report. For each finding, extract:

  • Finding ID (auto-assign if not present)
  • Metric affected (LCP, INP, CLS, TTFB, Bundle, Caching, etc.)
  • Severity (CRITICAL, HIGH, MEDIUM, LOW)
  • Estimated impact (e.g. "−800 ms LCP", "+0.05 CLS savings")
  • Owner (Engineering, Design, DevOps, Content)
  • Effort (XS, S, M, L)
  • Platform specificity (all platforms, Sitecore only, etc.)

Group findings by:

  1. Immediate / Quick Wins — CRITICAL severity or XS/S effort with HIGH impact
  2. Sprint 1 — HIGH severity findings
  3. Sprint 2 — MEDIUM severity findings
  4. Backlog — LOW severity or purely aspirational improvements

Phase 2 — Effort Calibration

Map effort estimates to story points (Fibonacci):

Effort Story points Description
XS 1 < 30 minutes; single-line config or attribute change
S 2–3 < 2 hours; single file or component change
M 5 Half-day; pipeline config or multi-file change
L 8 Full sprint story; architectural change or refactor
XL 13 Epic-level; requires design + engineering + DevOps

Phase 3 — Generate Work Items

Azure DevOps Format

Output one User Story per HIGH/CRITICAL finding and one Task per MEDIUM finding. Group related stories under a Feature titled "Performance Improvements — [Sprint/Quarter]".

User Story template:

Title: [PERF] <short finding title>

As a site visitor on <platform/page>,
I want <page/component> to load <metric> within <threshold>,
So that I experience fast page interactions that meet Google's Core Web Vitals.

Acceptance Criteria:
- [ ] LCP for <page> measures ≤ 2.5 s at p75 (Lighthouse mobile, 3 runs averaged)
- [ ] CLS for <page> measures ≤ 0.1 (Lighthouse mobile)
- [ ] Lighthouse CI passes in PR build without manual override
- [ ] <specific fix applied — e.g. "Hero image serves WebP from CDN with fetchpriority=high">
- [ ] <no regression — related metrics remain within budget>

Notes:
- Finding ID: PERF-<N>
- Metric: <LCP | INP | CLS | TTFB | Bundle>
- Estimated impact: <delta>
- Reference: <audit report filename>

Story Points: <N>
Tags: performance, core-web-vitals, <platform>

Task template (for MEDIUM/LOW findings):

Title: [PERF-TASK] <short finding title>

Description:
<finding description from audit report>

Fix:
<exact code change or config update from audit finding>

Definition of Done:
- [ ] Fix applied
- [ ] Lighthouse CI green on PR
- [ ] No CWV regressions introduced

Story Points: <N>
Parent Story: [link to parent User Story if part of a group]

Jira Format

Output in a Markdown table that can be pasted into Jira's bulk-import CSV format, or formatted for Jira's REST API.

Epic: "Performance Optimisation — [project/quarter]" Story type: Story for HIGH/CRITICAL, Sub-task for MEDIUM/LOW

Epic: Performance Optimisation
  Story: [PERF] Eliminate render-blocking resources (HIGH)
    - Subtask: Add `defer` to third-party scripts
    - Subtask: Implement critical CSS inlining
  Story: [PERF] Optimise hero image delivery (HIGH)
    - Subtask: Configure Next.js Image with fetchpriority
    - Subtask: Add WebP rule to Sitecore Media Library CDN
  Story: [PERF] Fix CLS from unsized images (MEDIUM)
  Story: [PERF] Configure Azure Front Door caching rules (HIGH)
    - Subtask: Set Cache-Control headers on content pages
    - Subtask: Add immutable rule for versioned static assets

Markdown Format

Output a prioritised backlog table for review in pull requests or Confluence:

## Performance Backlog — [Project Name]

Generated from: [audit report filename] | [date]

### Immediate / Quick Wins (do before next sprint)
| ID | Finding | Metric | Impact | Owner | Effort | Sprint |
|---|---|---|---|---|---|---|
| PERF-01 | Add `defer` to jQuery include | FCP/LCP | −320 ms FCP | Engineering | XS | Now |
| PERF-02 | Add `width`/`height` to hero `<img>` | CLS | −0.12 CLS | Engineering | XS | Now |

### Sprint 1 — High Impact
| ID | Finding | Metric | Impact | Owner | Effort | Sprint |
|---|---|---|---|---|---|---|
| PERF-03 | Configure Azure Front Door cache for content pages | TTFB/LCP | −900 ms TTFB | DevOps | M | Sprint 1 |
| PERF-04 | Convert hero image to WebP + responsive srcset | LCP | −600 ms LCP | Engineering | S | Sprint 1 |

### Sprint 2 — Medium Impact
| ID | Finding | Metric | Impact | Owner | Effort | Sprint |
|---|---|---|---|---|---|---|
| PERF-05 | Inline critical CSS for homepage | FCP | −180 ms FCP | Engineering | M | Sprint 2 |

### Governance (CI/CD — do once, permanent benefit)
| ID | Task | Benefit | Owner | Effort |
|---|---|---|---|---|
| PERF-GOV-01 | Set up Lighthouse CI in GitHub Actions | Prevents future regressions | Engineering | M |
| PERF-GOV-02 | Add bundle size budget (`size-limit`) | Prevents JS bloat | Engineering | S |

Phase 4 — Output Summary

After generating the backlog, output a brief summary:

## Backlog Summary

**Total findings:** N
**CRITICAL:** N | **HIGH:** N | **MEDIUM:** N | **LOW:** N

**Estimated total effort:** ~N story points across N sprints

**Projected CWV improvement (if all HIGH+ fixed):**
- LCP: [current] → [projected] (PASS/FAIL)
- INP: [current] → [projected] (PASS/FAIL)
- CLS: [current] → [projected] (PASS/FAIL)

**Recommended sprint allocation:**
- Sprint 1: N stories, N points — targets LCP and TTFB
- Sprint 2: N stories, N points — targets CLS and bundle size
- Ongoing: CI governance (Lighthouse CI, bundle size budget)

Quality Checklist

  • Every finding from the audit report maps to at least one work item
  • All CRITICAL findings are in Sprint 1 or "Immediate"
  • Each work item has measurable acceptance criteria referencing a CWV threshold
  • Owner assigned to every item (Engineering/Design/DevOps/Content)
  • Governance epic included (Lighthouse CI setup) if not already in place
  • Projected CWV improvement calculated if sufficient data in the audit report
Technical details
Plugin
performance-auditor
Author
Dakota Smith
Skill name
perf-backlog-builder
Maturity
experimental
Security review
No security review
Owners
  • engineering
Argument hint
<audit-report-path> [--format ado|jira|markdown] [--sprints <n>] [--team <team-name>]
Allowed tools
  • Read
  • Write
  • Bash(cat *)
  • Bash(ls *)
Source
plugins/performance-auditor/skills/perf-backlog-builder/SKILL.md

Edit this skill on GitHub — opens GitHub's web editor. No git knowledge needed.