Borrowing it
Nothing to install: this file belongs to kikin81/nubecita. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/kikin81/nubecita/main/.claude/skills/run-startup-bench/SKILL.mdgit clone --depth 1 https://github.com/kikin81/nubecitaWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kikin81/nubecita/run-startup-bench)<a href="https://agentmods.dev/skills/kikin81/nubecita/run-startup-bench"><img src="https://agentmods.dev/badge/skills/kikin81/nubecita/run-startup-bench/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kikin81/nubecita/run-startup-bench"><img src="https://agentmods.dev/badge/skills/kikin81/nubecita/run-startup-bench.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00071 | $0.02143 |
| Opus 5 | $0.00036 | $0.01071 |
| Sonnet 5 | $0.00014 | $0.00429 |
| Haiku 4.5 | $0.00007 | $0.00214 |
Grade A, and why
run-startup-bench 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 11d ago.
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 — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Project-local skill for running :benchmark's StartupBenchmark and reading the comparison cleanly. Companion to benchmark/README.md (the long-form workflow doc).
Announce at start: "I'm using the run-startup-bench skill to run the macrobench and compare cells."
When to use
Trigger on any of:
- "Run the startup bench" / "run macrobench" / "kick off the startup tests"
- "Measure the profile's impact" / "is the baseline profile working"
- "Compare bench results" / "is COLD startup faster than before"
- "Regenerate the baseline profile" (the gen task — different command but same device hygiene)
If the user wants FeedScrollBenchmark instead, do NOT use this skill — that bench needs the asset-backed fake (nubecita-crmi.5 epic) which isn't shipped yet.
Pre-flight (always)
The two most expensive failure modes of :benchmark runs eat ~5–15 min each. Run all four before any bench command:
-
Device connected, single instance.
adb devices -lIf a Pixel Tablet or other secondary device shows up over TCP (Wi-Fi ADB auto-pair),
adb disconnect <ip:port>it — multi-device confuses gradle's picker and the next test fails with "No connected devices!". -
Stay-awake on USB (prevents mid-run sleep that drops ADB):
adb -s <serial> shell settings put global stay_on_while_plugged_in 7 -
Pin the serial in the gradle invocation:
-Pandroid.testInstrumentationRunnerArguments.androidx.benchmark.targetDeviceSerial=<serial> -
Only the
productiongenerator needs sign-in. The gen task targets:app'sbenchflavor by default (fake repos, fakeSignedIn, mock feed) — no sign-in, never routes to Login, but it produces a profile that's missing ~13k real network/crypto/serialization rules, so don't ship it. For the representative, shippable profile pass-PbaselineProfileEnvironment=production: that drives the real signed-in cold start, so first installproductionNonMinifiedReleaseand sign in (it fails fast if Splash routes to Login). The bench (not the generator) wipes app data each run, so its sign-in never needs to persist.
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.
- 11d ago First seen · 164 lines · 71 tokens per session scan A 780ca8dd9723
run-startup-bench is a skill published in the GitHub repository kikin81/nubecita (5 stars, last pushed today), licensed MIT. It adds 71 tokens to every session and 2,143 once invoked, about $0.0004 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
android_ui_verification
Automated end-to-end UI testing and verification on an Android Emulator using ADB.
agent-device
Automates Apple-platform apps (iOS, tvOS, macOS), Android devices, and Amazon Vega OS TV apps in Vega Virtual Devices. Use when navigating apps, taking snapshots/screenshots where supported, driving TV remotes, tapping, typing, scrolling, extracting UI info, collecting evidence, or planning agent-device CLI commands.
dogfood
Systematically explore and test a mobile app on iOS/Android with agent-device to find bugs, UX issues, and other problems. Use when asked to dogfood, QA, exploratory test, find issues, bug hunt, or test this app on mobile.
ios-simulator
Verify and debug native, React Native, Expo, or Flutter apps on an iOS Simulator with agent-device. Use when an agent needs to launch an app, inspect its live UI, tap, type, scroll, validate a code change, collect failure evidence, or reproduce a workflow on an iPhone or iPad Simulator.
solopi-ai
A command-line framework for testing Android apps and devices with SoloPi, including on-device or cloud AI decision models. It manages devices, test cases, recorded interactions, replays, performance history, and evidence.
eas-simulator
EAS service (paid). Run and control a user's app on a remote iOS/Android simulator hosted on EAS cloud. Read before running any eas simulator: commands - it has the current syntax for this experimental API. Use whenever the user needs a simulator they can't run locally - 'run my app on a cloud simulator', 'use eas…