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 skills/millionco/debug-agent/web-performancenpx skills add millionco/debug-agent --skill web-performancegit clone --depth 1 https://github.com/millionco/debug-agentWhat 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 | $0.00057 | $0.03877 |
| Opus 5 | $0.00028 | $0.01938 |
| Sonnet 5 | $0.00011 | $0.00775 |
| Haiku 4.5 | $0.00006 | $0.00388 |
Grade A, and why
web-performance scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
To clear the log file via HTTP without deleting/recreating it: `curl -X DELETE <endpoint>` returns `{"ok":true,"cleared":true}`. How it starts
The opening of the file, as written. The whole thing — 257 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Web Performance
Overview
Browser performance debugging via PerformanceObserver, with LoAF (long-animation-frame) as the primary signal. LoAF is the only entry type that, in one record, attributes a slow frame to a specific sourceURL + sourceFunctionName + sourceCharPosition + invokerType, with per-script forcedStyleAndLayoutDuration (sync reflow), pauseDuration (sync XHR / alert), and blockingDuration. Code inspection and performance.now() cannot reach this. Start with LoAFs, conclude from LoAFs.
When to use
Symptoms:
- Jank, dropped frames, janky scroll/swipe, complaints about frame rate
- Slow click / keypress / touch response, "unresponsive" complaints, poor INP
- Slow LCP, layout shifts (CLS)
- Animation stutter, transition jank, expensive renders during interaction
Do NOT use for:
- Backend / non-browser perf — use raw NDJSON file appends from your server runtime
- Memory leaks, bundle-size regressions — heap snapshots / bundle analyzers
- Logic bugs unrelated to timing — use raw fetch instrumentation
Core pattern
Before — manual performance.now() (wrong):
const t0 = performance.now();
drawSeries(data);
console.log("drawSeries took", performance.now() - t0);
Tells you a number. Doesn't tell you the function caused a long frame, what scheduled it, or whether it forced sync layout. Requires you to already suspect drawSeries.
After — LoAF observer (right):
new PerformanceObserver((list) => {
for (const loaf of list.getEntries()) send(loaf);
}).observe({ type: "long-animation-frame", buffered: true });
Reports every frame > 50ms across the whole page, with scripts[].sourceURL + sourceFunctionName + sourceCharPosition + invokerType + forcedStyleAndLayoutDuration for each script that ran in the frame. You don't need to know where the bug is in advance.
Workflow
- Generate 3-5 hypotheses about what's slow and where.
- Start the logging server (Implementation → STEP 0).
- Inject the LoAF observer as the first script in
<head>or top of SPA entry. - Reproduce — automate via Playwright/Puppeteer if possible; otherwise give numbered steps and ask the user to confirm in their UI (do NOT ask them to type "done").
- Clear the log file before each run via the deletion tool (NOT
rm). - Analyze LoAFs first; consult secondary signals only if LoAF is silent. Mark hypotheses CONFIRMED / REJECTED / INCONCLUSIVE with cited entries.
- Fix only with 100% confidence. Keep instrumentation in place; tag post-fix runs with
runId="post-fix". - Verify by re-running and comparing before/after LoAFs with cited lines. If failed, revert rejected-hypothesis code (keep instrumentation), generate new hypotheses, iterate.
- Cleanup — remove the
#region debug logblock only after verified success + explicit user confirmation.
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 · 257 lines · 57 tokens per session scan A 0cc46a9960ea
web-performance is a skill published in the GitHub repository millionco/debug-agent (302 stars, last pushed 3mo ago), licensed MIT. It adds 57 tokens to every session and 3,877 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
design-taste-frontend-v1
The original v1 taste-skill, preserved for projects depending on its exact behavior. The current default is design-taste-frontend (v2 experimental), which is a substantial rewrite. Use this v1 install name only if you need exact backward compatibility.
brandkit
Premium brand-kit image generation skill for creating high-end brand-guidelines boards, logo systems, identity decks, and visual-world presentations. Trained for minimalist, cinematic, editorial, dark-tech, luxury, cultural, security, gaming, developer-tool, and consumer-app brand systems. Optimized for intentional…
high-end-visual-design
Teaches the AI to design like a high-end agency. Defines the exact fonts, spacing, shadows, card structures, and animations that make a website feel expensive. Blocks all the common defaults that make AI designs look cheap or generic.
gpt-taste
Elite UX/UI & Advanced GSAP Motion Engineer. Enforces Python-driven true randomization for layout variance, strict AIDA page structure, wide editorial typography (bans 6-line wraps), gapless bento grids, strict GSAP ScrollTriggers (pinning, stacking, scrubbing), inline micro-images, and massive section spacing.
iflytek-hyper-tts
Use when user asks to synthesize speech, convert text to audio, or read text aloud. 讯飞超拟人语音合成 - 支持文本转语音、语音合成（发音人/语速/语调/音量/输出格式）。大模型语音合成技能。语音合成, 文字转语音, 超拟人, TTS.
iflytek-voiceclone-tts
Use when user asks to clone a voice, train a custom voice model, or synthesize speech with a cloned voice. iFlytek Voice Clone tts(声音复刻) — train a custom voice model from audio samples and synthesize speech with the cloned voice. Supports the full workflow: get training text → create task → upload audio → submit…