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 skills add niki914/zafiro --skill phone-usegit clone --depth 1 https://github.com/niki914/zafiroWrote 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/niki914/zafiro/phone-use)<a href="https://agentmods.dev/skills/niki914/zafiro/phone-use"><img src="https://agentmods.dev/badge/skills/niki914/zafiro/phone-use/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/niki914/zafiro/phone-use"><img src="https://agentmods.dev/badge/skills/niki914/zafiro/phone-use.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00076 | $0.02929 |
| Opus 5 | $0.00038 | $0.01465 |
| Sonnet 5 | $0.00015 | $0.00586 |
| Haiku 4.5 | $0.00008 | $0.00293 |
Grade A, and why
Phone Use 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 10d 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 — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Overview
Phone Use gives you the ability to see and interact with the Android device screen. It uses Android AccessibilityService (with root-assisted auto-setup) to read the UI tree and simulate user actions.
This is the ONLY way to control the device — do NOT attempt to use non-existent APIs or platform tools.
When to load: any GUI task. App launching, tapping, typing, scrolling, swiping, navigation keys, screen reading. Even when the user doesn't say "on the phone" — infer it from the task.
When NOT to retry: non-native apps (Flutter, Unity, WebView, games). If screen_operation_accessibility(operation: "read") returns empty/root-only, stop and report immediately.
| Tool | Purpose |
|---|---|
screen_operation_accessibility |
Read screen tree, tap/long-click/scroll/set_text on nodes by token, search nodes — all via accessibility service |
screen_operation_shell |
Tap/long-click/swipe/key by screen coordinates via shell. FALLBACK only — coordinates MUST come from the most recently returned screen tree |
Two auxiliary tools for app discovery and launching:
| Tool | Purpose |
|---|---|
find_installed_apps |
Find installed apps by name or package name |
launch_app |
Launch an app by package name or fuzzy app name |
Tool Reference
See each tool's own description for full parameter docs, operation lists, YAML field glossary, and key codes. This section covers cross-tool usage rules and constraints.
screen_operation_accessibility
Primary tool for reading the screen and interacting with labeled UI nodes. The YAML header contains the snapshot version. Each node has an i (index) field. To target a node, construct the token by joining the snapshot version with the node's i, separated by _: e.g. header version: "a3f2c91e7b40" + node {i: 42, ...} → token "a3f2c91e7b40_42". Every successful write operation (tap, long_click, scroll_forward, scroll_backward, set_text) auto-returns the updated YAML tree — no separate read needed.
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.
- 10d ago First seen · 178 lines · 76 tokens per session scan A 74de1bde9c27
Phone Use is a skill published in the GitHub repository niki914/zafiro (153 stars, last pushed today), licensed MIT. It adds 76 tokens to every session and 2,929 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 2026-08-30.
Other skills, from other repositories
kotlin-expert
Expert-level Kotlin development, Android, coroutines, and multiplatform. Use when the user mentions Android, coroutines, multiplatform, or JVM, or when the task involves Kotlin Fundamentals, Android Development, Kotlin Multiplatform, or Kotlin Style.
dagger-hilt-expert
Expert guidance for Dagger and Hilt dependency injection in Android. Use when implementing DI, creating modules, configuring scopes, optimizing performance, testing with Hilt, setting up multi-module architecture, using assisted injection, or debugging DI issues. Triggers on "dagger", "hilt", "dependency injection"…
android-startup-optimizer
Expert Android app startup performance optimization skill. Use when analyzing or improving Android app cold start, warm start, or hot start times. Applies to projects using Jetpack Compose, Hilt/Dagger, ContentProviders, ad SDKs, or any Android initialization bottlenecks. Triggers on phrases like "startup time", "app…
android-amplitude-sdk-expert
Expert guide for Amplitude Android SDK integration (Kotlin/Java). Use when tracking events, configuring autocapture, managing user identity, configuring batching or EU residency, implementing custom Enrichment/Destination plugins, migrating from legacy com.amplitude:android-sdk to modern…
android-leak-expert
Expert guide for detecting, profiling, and resolving memory leaks in Android apps (Java, Kotlin, Compose, JNI). Use when analyzing memory usage, profiling heap dumps, fixing out-of-memory (OOM) issues, checking Compose lifecycle flow collections, resolving WebView/Handler leaks, or configuring LeakCanary. Triggers…
compose-component-expert
Expert guide for designing, building, and optimizing reusable Jetpack Compose components. Applies 2026 best practices, Style API, slot APIs, Modifier.Node, strong skipping mode, recomposition optimization, and Detekt/Ktlint rules validation. Trigger this skill whenever the user asks about Jetpack Compose component…