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/purewhiter/mobilegym/agents-mdgit clone --depth 1 https://github.com/Purewhiter/mobilegymWrote 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/instructions/purewhiter/mobilegym/agents-md)<a href="https://agentmods.dev/instructions/purewhiter/mobilegym/agents-md"><img src="https://agentmods.dev/badge/instructions/purewhiter/mobilegym/agents-md.svg" alt="Measured on agentmods" 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 | $0.06877 | $0.06877 |
| Opus 5 | $0.03438 | $0.03438 |
| Sonnet 5 | $0.01375 | $0.01375 |
| Haiku 4.5 | $0.00688 | $0.00688 |
Grade A, and why
mobilegym AGENTS.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 5d 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 — 328 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
This file provides guidance to Code Agents when working with code in this repository. Respond in whatever language the user writes to you in — if they ask in Chinese, reply in Chinese; if they ask in English, reply in English.
Project Overview
mobile-gym is a simulated Mobile environment (Android-like) built with React + Vite + TypeScript + Tailwind CSS v4. It serves as a training and benchmarking platform for mobile GUI Agents. The simulator runs in a browser and exposes JavaScript APIs (__SIM__, __OS__, __SIM_INPUT__, __SIM_QUERY__, __SIM_TIME__, __SIM_LOCATION__, __SIM_FS__) for task management, trajectory data synthesis, and benchmark orchestration. The Agent only sees screenshots.
User-facing documentation under docs/ and bench_env/docs/ is in English. New user-visible App text must also follow the i18n resource contract: keep it in res/strings.ts with localized overrides such as res/strings.en.ts, and consume it through useAppStrings / resolveAppStrings instead of scattering Chinese or English literals through JSX. See docs/platform/tooling/i18n.md and docs/platform/app/resources.md.
Type-checking strategy
- Small changes (touching a few files, styling tweaks, data updates) — no need to run
tsc --noEmit; rely on the IDE's live checker - Large changes — run
npx tsc --noEmitonce on completion to confirm there are no type errors
ESLint
npm run lint # Lint runtime code under os/ and apps/
Current rule: bare Date.now() and any form of new Date(...) (including parameterised forms — they must go through TimeService) are forbidden. Config lives in eslint.config.js.
Navigation Artifact Generation (run after modifying navigation declarations)
# One-shot: consistency check + schema nav graph + action tasks
node scripts/build_nav_artifacts.mjs <AppName>
# With data graph generation
node scripts/build_nav_artifacts.mjs <AppName> --data data/index.ts
# Skip tasks, only update graphs
node scripts/build_nav_artifacts.mjs <AppName> --skip-tasks
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.
- 5d ago First seen · 328 lines · 6,877 tokens per session scan A a94b3d91236c
mobilegym AGENTS.md is an instructions file published in the GitHub repository Purewhiter/mobilegym (784 stars, last pushed 7d ago), licensed Apache-2.0. It adds 6,877 tokens to every session, about $0.0344 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 instructions, from other repositories
Gym AGENTS.md
AGENTS.md instructions for NVIDIA-NeMo/Gym, covering agents.md, quality bar, what this is, architecture and creating environments.
Gym CLAUDE.md
Claude Code instructions for NVIDIA-NeMo/Gym, a project described as: Evaluate and improve models and agents using environments.
react-doctor AGENTS.md
AGENTS.md instructions for millionco/react-doctor, covering general rules, symbol search & deduplication (truffler), package layout, effect v4 conventions and imports.
agent-belt AGENTS.md
AGENTS.md instructions for jfrog/agent-belt, covering agent-belt - guide for ai coding agents, 1. what this is, 2. setup & first verification, 3. where to start (by task) and 4. design principles.
agentic-mobile-blueprint AGENTS.md
Instructions for Ampli-Group/agentic-mobile-blueprint, covering agent context, task workflow, 1. check environment, 2. implement and 3. verify.
tamagui CLAUDE.md
Claude Code instructions for tamagui/tamagui, a project described as: Style React fast with 100% parity on React Native, an optional UI kit, and optimizing compiler.