Index of every page on this site: llms.txt
All skills
elithrar avatar

/web-perf

@eb4f994
by Matt Silverlockelithrar/dotfiles226 stars
16
Run it yourself

Diagnose page-load and interaction performance using browser traces, network evidence, and source code. Use for site-speed, Lighthouse, Core Web Vitals, or layout-shift investigations; a general accessibility audit is separate.

Use this Skill: https://skilld.dev/gh/elithrar/dotfiles/web-perf

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈59 tokens for metadata: the name and description. ≈655 when used: this file. ≈798 more on demand in 2 files.

Description uses 3.0% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈238 characters, or 3.0%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Web performance investigation

Identify the target page, reported symptom, and relevant device conditions. For audit requests, report findings. For optimization or fix requests, implement the supported changes and validate their effect within the requested scope.

Choose the evidence path

Inspect the available browser and profiling tools before calling them. Prefer Chrome DevTools traces when available; read references/devtools.md for trace and network investigation. Use the installed tool schemas rather than assuming example parameters match the current version.

If that MCP server is unavailable, use existing trace or Lighthouse artifacts, another available profiler, or relevant source code. Complete the analysis those sources support and state what could not be measured. Missing profiling tools do not justify stopping source inspection or forcing an MCP installation.

Diagnose

Record the page/build, viewport, device or emulation, network/CPU throttling, cache state, and interactions relevant to the result. Reloading alone does not establish a cold cache. Keep mobile and desktop observations distinct.

Separate field Core Web Vitals from laboratory measurements. A load-only trace does not establish INP; TBT is a diagnostic lab metric, not a substitute INP measurement. Use the current metric definitions when assigning ratings, and label missing measurements as unavailable.

Follow the bottleneck from trace insight to request, element, or source path. Inspect the relevant LCP phases, layout-shift participants, long tasks, late resource discovery, or cache headers. Confirm ownership and use before recommending removal, especially for third-party scripts or resources used only in later interactions. Zero estimated savings is not evidence of a material problem, but one trace cannot prove a resource is never useful.

Validate and report

Prioritize findings by observed user impact and confidence. Distinguish measured values, tool estimates, and unmeasured hypotheses; avoid promising additive savings from overlapping insights.

For a fix, compare the affected behavior under equivalent conditions. Repeat measurements when variance or the size of the claimed improvement warrants it. Broaden testing only for an unresolved regression risk or a project gate, then finish.

Report the measured symptom, its supported cause, the concrete fix or recommendation, and the relevant verification limit. Add a metric table when several measurements help comparison. Do not expand a performance request into a mandatory accessibility or whole-codebase audit.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

2 warnings2mo4 checks · Risk SAFE
  • Gen Agent Trust Hub2mo

    The skill is a diagnostic tool for web performance that analyzes browser traces and network evidence. It is generally safe but carries a minor risk of indirect prompt injection because it processes untrusted content from external websites. It also references official documentation from a well-known service.

  • Socket2mo

    No alerts

  • Snyk2mo

    Risk: MEDIUM · 1 issue

  • Runlayer9mo

    1/1 file flagged

Signed by skilld at eb4f994. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 3 months ago

README badge

README badge for elithrar/dotfiles/web-perf