Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/dev-lou/pixelpilot/performance-auditorgit clone --depth 1 https://github.com/dev-lou/PixelPilotWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03104 | $0.03104 |
| Opus 5 | $0.01552 | $0.01552 |
| Sonnet 5 | $0.00621 | $0.00621 |
| Haiku 4.5 | $0.00310 | $0.00310 |
Grade A, and why
PixelPilot performance-auditor.instructions.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 364 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance Auditor 2026
Run → capture → score → fix. Automates Lighthouse audits via MCP (headless Chrome), reports CWV, and maps every failing audit to a concrete, token-aware fix.
CORE WEB VITALS TARGETS (2026)
| Metric | Good | Needs Improvement | Poor |
|---|---|---|---|
| LCP Largest Contentful Paint | ≤ 2.5 s | 2.5–4.0 s | > 4.0 s |
| INP Interaction to Next Paint | ≤ 200 ms | 200–500 ms | > 500 ms |
| CLS Cumulative Layout Shift | ≤ 0.1 | 0.1–0.25 | > 0.25 |
| FCP First Contentful Paint | ≤ 1.8 s | 1.8–3.0 s | > 3.0 s |
| TTFB Time to First Byte | ≤ 800 ms | 800–1800 ms | > 1800 ms |
| TBT Total Blocking Time | ≤ 200 ms | 200–600 ms | > 600 ms |
Score targets:
- Performance: ≥ 90
- Accessibility: ≥ 95
- Best Practices: ≥ 95
- SEO: ≥ 90
STEP 1 — CHOOSE AUDIT METHOD
Method A — Chrome DevTools MCP (preferred for local dev)
// Navigate to target
await browser_navigate({ url: TARGET_URL });
// Wait for page to fully load
await browser_evaluate({ script: `
await new Promise(resolve => {
if (document.readyState === 'complete') return resolve();
window.addEventListener('load', resolve);
});
await new Promise(r => setTimeout(r, 1000));
` });
// Collect Web Vitals via PerformanceObserver
const vitals = await browser_evaluate({ script: `
const nav = performance.getEntriesByType('navigation')[0];
const paint = performance.getEntriesByType('paint');
const lcp = await new Promise(res => {
new PerformanceObserver(list => {
const entries = list.getEntries();
res(entries[entries.length - 1].startTime);
}).observe({ type: 'largest-contentful-paint', buffered: true });
setTimeout(() => res(null), 3000);
});
return {
TTFB: nav?.responseStart - nav?.requestStart,
FCP: paint.find(e => e.name === 'first-contentful-paint')?.startTime,
LCP: lcp,
TBT: nav?.domComplete - nav?.domInteractive,
domSize: document.querySelectorAll('*').length,
scripts: performance.getEntriesByType('resource')
.filter(r => r.initiatorType === 'script')
.map(r => ({ url: r.name.split('/').pop(), size: r.transferSize, duration: r.duration }))
.sort((a,b) => b.size - a.size)
.slice(0, 10),
images: performance.getEntriesByType('resource')
.filter(r => r.initiatorType === 'img')
.map(r => ({ url: r.name.split('/').pop(), size: r.transferSize }))
.sort((a,b) => b.size - a.size)
.slice(0, 10),
thirdParty: performance.getEntriesByType('resource')
.filter(r => !r.name.includes(location.hostname))
.reduce((acc, r) => acc + r.transferSize, 0),
};
` });
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 364 lines · 3,104 tokens per session scan A b4fb6892ef80
PixelPilot performance-auditor.instructions.md is an instructions file published in the GitHub repository dev-lou/PixelPilot (2 stars, last pushed 4mo ago), licensed MIT. It adds 3,104 tokens to every session, about $0.0155 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
arai CLAUDE.md
Instructions for taniwhaai/arai, covering claude.md — arai, commands, architecture, prompt-collector module (src/promptcollector.rs) and extending the match pipeline.
healthcare-agents AGENTS.md
Instructions for ajhcs/healthcare-agents, covering healthcare agents repository instructions, repository map, git workflow and self-improvement loop.
arai AGENTS.md
Instructions for taniwhaai/arai, covering agents.md — arai, core discipline (non-negotiable), taniwha / subagent rules, work style and tool usage.
shenzjd-skills CLAUDE.md
Claude Code instructions for wu529778790/shenzjd-skills, covering claude.md, project overview, commands, a single skill's unit test and architecture.
ui-ux-design-pro-skill CLAUDE.md
Instructions for saifyxpro/ui-ux-design-pro-skill, covering project overview, cli commands, full design system generation (markdown output), search across all databases (sub-50ms) and audit ui code quality.
syncai copilot-instructions.md
Instructions for dmtrkzntsv/syncai, covering claude.md, build & run, architecture, lifecycle and identify is the routing decision.