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/kouroshez/coding-os/performancenpx skills add kouroshez/coding-os --skill performancegit clone --depth 1 https://github.com/kouroshez/coding-osWhat 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 | $0.00090 | $0.03517 |
| Opus 5 | $0.00045 | $0.01758 |
| Sonnet 5 | $0.00018 | $0.00703 |
| Haiku 4.5 | $0.00009 | $0.00352 |
Grade A, and why
performance 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 — 306 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance — Web + Backend + Mobile
A practical playbook for measuring, profiling, and improving performance across the project's stack. Covers Web Vitals (Google's user-experience metrics), mobile perf budgets (FPS / TTI / memory), backend latency (P95 / P99 / saturation), with concrete tools per surface.
When to Use This Skill
- Adding a perf budget to a new feature.
- Investigating a Lighthouse / Web Vitals regression.
- Mobile users report jank, slow app launch, battery drain.
- Backend P95 latency exceeds SLO.
- Bundle size jumped 30% after a release.
- Lists scroll poorly; animations stutter.
- Choosing between perf tools (when each pays off).
The Three Performance Surfaces
| Surface | Primary metrics | Tools |
|---|---|---|
| Backend | P50 / P95 / P99 latency, throughput, error rate, saturation (CPU / mem / IO) | OpenTelemetry, Datadog APM, pprof (Go), py-spy (Python), pgbadger (Postgres) |
| Web frontend | LCP, INP, CLS (Core Web Vitals), TTI, TBT, bundle size | Lighthouse, Chrome DevTools Performance, WebPageTest, web-vitals JS lib |
| Mobile | FPS (60/120), TTI, memory, bundle size, native render time | Flipper, React DevTools Profiler, Hermes Sampling Profiler, Xcode Instruments, Android Profiler, FlashList Recorder |
Each has its own measurement approach; conflate them at your peril.
Measurement First, Optimization Second
The cardinal rule: don't optimize without numbers. The biggest performance regressions in the wild come from devs "fixing" something that wasn't slow.
Workflow:
- Establish a budget (e.g., LCP < 2.5s P75, list scroll 60 FPS, P95 < 200ms).
- Measure on real conditions (real device, real network, real data volume).
- Identify the bottleneck via profiling.
- Fix the bottleneck.
- Re-measure to confirm improvement.
- Pin a regression test if possible (CI bundle-size gate, k6 load test).
Web Vitals (2026 — current Google ranking signals)
| Metric | What it measures | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP (Largest Contentful Paint) | Time until largest element renders | ≤ 2.5s | 2.5–4.0s | > 4.0s |
| INP (Interaction to Next Paint) | Worst interaction latency in session — replaced FID March 2024 | ≤ 200ms | 200–500ms | > 500ms |
| CLS (Cumulative Layout Shift) | Visual stability over the page lifetime | ≤ 0.1 | 0.1–0.25 | > 0.25 |
| TTFB (Time to First Byte) | Server response start | ≤ 0.8s | 0.8–1.8s | > 1.8s |
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 306 lines · 90 tokens per session scan A 390bf52bbcf1
performance is a skill published in the GitHub repository kouroshez/coding-os (6 stars, last pushed yesterday), licensed Apache-2.0. It adds 90 tokens to every session and 3,517 once invoked, about $0.0005 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 skills, from other repositories
design-build
Build UI from this repository's primitives and ramps in a deterministic order, then prove the rendered result with the matching instruments and gates.
raytsystem-watch
Inspect video, audio, or supplied transcripts through raytsystem Tool Hub and return evidence-bound speech, visual, OCR, action, transition, and timeline findings. Use for /watch, a YouTube/Loom/public Zoom/direct media URL, a local video or audio file, a transcript, or requests such as "watch this video", "analyze…
watch
Inspect video, audio, or supplied transcripts through raytsystem Tool Hub and return evidence-bound speech, visual, OCR, action, transition, and timeline findings. Use for /watch, a YouTube/Loom/public Zoom/direct media URL, a local video or audio file, a transcript, or requests such as "watch this video", "analyze…
ingest
Use when a raw source file needs to be ingested into the knowledge graph.
audio-transcriber-knowledge-graph
Native knowledge-graph ingestion of Whisper transcripts on the audio-transcriber MCP server — transcribe an audio/video file and push it into the epistemic-graph in one call: the raw audio as a shared :MediaAsset blob, the transcript text as a :Document, and each Whisper segment as a :TranscriptSegment node linked…
t2i
Use the t2i CLI to generate AI images from text prompts via Microsoft Foundry providers (FLUX.2, MAI-Image-2). Activate when the user asks to generate images, automate image creation in scripts, or set up image generation for CI/CD.