Resources

Master prompts

Improve website performance — master prompt

Find the dominant loading or interaction cost and prove whether a focused change helps.

CodexClaude CodeCursor
Pristren originalFree1.0.0 · original authored template; documentation reviewed 2026-11-17

A measured performance improvement with a reproducible baseline and no lost critical functionality.

Use this prompt after supplying a concrete project or task and the constraints that determine success. Select your AI tool in the workflow for host-specific setup. Review the prerequisites and choose only relevant resources; the prompt does not automatically include an entire directory.

Investigate the largest observed cost before changing dependencies or rewriting the page.

After execution, check the workflow’s acceptance criteria against observed behavior. Treat illustrative examples as examples, and record anything not actually verified. Project details stay in your browser when you tailor this template.

Make the prompt yours.

Master prompt

Complete project or task to copy your tailored prompt.

Improve performance for [Describe the project, audience, and useful result.] while respecting Use existing project conventions. State assumptions and verify the result..

Inspect the page, build output, and current loading behavior. Record the tested URL/build, device and network settings, relevant metrics, and the main user journey. Distinguish lab observations from field data. Do not promise a score or infer performance from source size alone.

Find the dominant observed cost: critical requests, oversized media, blocking scripts, unnecessary hydration, layout movement, or long interaction work. Explain the evidence and select one focused change with a plausible causal path. Keep the page’s important content, accessibility, analytics requirements, and business action intact. Prefer deferring optional media or reducing avoidable work over a broad rewrite.

Make the change and repeat the same measurements with comparable conditions. Complete the main user task, inspect reduced motion and narrow screens if relevant, and check the production build. Report before/after observations with run variability and any tradeoff. If the result is inconclusive, say so and propose the next measurement instead of claiming a win.

Execution environment: Codex. Follow applicable project instructions and verify the requested result.

Use it with your tool.

Documentation reviewedChecked 2026-11-17

Tool setup checked against official documentation. Pristren has not executed this workflow through this tool.

  1. Start Codex in the repository and read applicable AGENTS.md guidance. Include only the files this task needs.
  2. Investigate the largest observed cost before changing dependencies or rewriting the page.
  3. Use the acceptance checks below to assess the result; an agent completion message is not evidence that the task passed.
Official setup documentation

License and access: MIT for this Pristren-authored prompt or instruction text.

Part of a bigger picture.

Workflow

Improve website performance

Find the dominant loading or interaction cost and prove whether a focused change helps.

Prompt, setup & checksCodex / Claude Code / Cursor

Go a little deeper.

Sources and review notes