Sanyam
Back to Agentic AI

Agentic AI / Reusable artifact

Lighthouse Audit Skill

A Claude Code Skill that runs Lighthouse against a live URL, cross-references every finding against this repo real source code, and writes a prioritized, file-and-line fix plan instead of generic advice.

Skills

A Skill packages up a recurring workflow: instructions, plus any scripts it needs, so Claude Code runs it the same way every time instead of reinventing the approach. This one automates the "why is my site slow / scoring badly" request end to end.

What it does

  1. Runs Lighthouse twice against a URL: mobile (throttled) and desktop. The gap between the two numbers is itself a diagnostic signal.
  2. Summarizes the raw JSON reports (often 10,000+ lines each) with a bundled Python script instead of reading them directly, keeping the agent's context window for analysis, not raw data.
  3. Traces every finding back to the actual source in this codebase: a specific import, a gatsby-plugin-image misconfiguration, a render-blocking script, instead of repeating Lighthouse's own generic advice.
  4. Writes a structured fix plan, ordered by real-world severity, that another session (or a cheaper model) can execute later with zero extra context.
<!-- diagram: flow-lighthouse-audit -->
graph TD
    A[Run Lighthouse: mobile + desktop] --> B[Summarize report with bundled script]
    B --> C[Diagnose each finding against this repo's source]
    C --> D[Write prioritized fix plan]
    D --> E{Fix now or later?}
    E -- Now --> F[Work through findings by severity]
    E -- Later --> G[Plan file is the handoff artifact]

Why a skill instead of a one-off prompt

The same audit gets asked for repeatedly, by different people, months apart. A skill means the process doesn't have to be re-explained or re-discovered each time: run both form factors, summarize instead of reading raw, verify against source, write a structured plan. It's version-controlled like any other project asset, so improvements to the workflow benefit every future audit.

How to get it

This skill is maintained publicly in Agentry Labs, a framework-agnostic collection of reusable AI-agent components. Install it into any project with one command:

curl -sSL https://raw.githubusercontent.com/ersanyamarya/agentry-labs/main/scripts/install.sh | bash -s -- skills/lighthouse-audit

Then invoke it inside Claude Code:

claude -p "/skill lighthouse-audit"

The skill directory contains:

  • SKILL.md: the instructions Claude Code follows
  • scripts/summarize_report.py: condenses a raw Lighthouse JSON report into the ~100 lines that actually matter
  • evals/evals.json: test cases used to check the skill still behaves as expected after edits