This repository contains AI Agent skills for Appium automation, including environment setup, real-device configuration, and troubleshooting common failures.
Appium Skills is repository-first. Each SKILL.md is a focused workflow entry point that owns its selection boundary, inputs, workflow outline, safety rules, loading instructions, and completion criteria. Shared knowledge remains in top-level contexts/, and deterministic checks remain in top-level tools/, so multiple Skills can reuse them without duplicating procedures.
Fine-grained Context Assets are intentional when they represent independently selectable platform, driver, validation, security, or troubleshooting concerns. Generated inventories and validation reports are projections of this architecture, not sources of domain guidance.
Each workflow includes or delegates to a self-improvement prompt in the loaded context, reference, or procedure. After use, note missing, ambiguous, outdated, or retry-causing instructions and propose a focused update to the asset that owns the affected guidance.
| Skill | Description |
|---|---|
| setup | Selects and coordinates the exact driver-specific setup Skills requested by the user |
| setup-uiautomator2 | Prepares and verifies Appium UiAutomator2 and Android prerequisites |
| setup-espresso | Prepares and verifies Appium Espresso and Android prerequisites |
| setup-chromium | Prepares and verifies Appium Chromium and a desktop Chromium browser |
| setup-gecko | Prepares and verifies Appium Gecko and Firefox |
| setup-mac2 | Prepares and verifies Appium Mac2 on macOS |
| setup-safari | Prepares and verifies Appium Safari on macOS |
| setup-xcuitest | Prepares and verifies shared Appium XCUITest prerequisites and continues to real-device configuration when applicable |
| prepare-development-environment | Prepares this repository for local development and validation by checking Git, Node, npm, helper scripts, Skill structure, repository-local paths, and JavaScript syntax |
| appium-troubleshooting | Diagnoses implemented UiAutomator2 startup, transport, and locator failures or XCUITest WDA, startup, device-state, and element-lookup failures, re-checking the smallest reproduction after each fix |
| xcuitest-real-device-config | Continues from XCUITest setup to prepare a real iOS/tvOS device |
- Run selected Skills one at a time in the order stated by the owning Skill body and applicable handoff conditions; Renma declaration order is not execution order.
- If output appears incomplete, rerun only the affected step and collect logs from that step.
- Treat Appium doctor as the source of truth for pass/fail (
0 required fixes needed). - For FFmpeg-dependent capabilities, use
contexts/tools/ffmpeg/setup.mdonly when explicitly requested. - For bundletool-dependent capabilities, use
contexts/platform/android/bundletool.mdonly when explicitly requested. - For supported UiAutomator2 or XCUITest startup, WDA, transport, device-state, or locator failures after setup is complete, use
appium-troubleshootingin a single driver path. - Repository authoring analyzers are maintainer-side tools and are not completion gates for Appium workflows.
Dedicated troubleshooting routes for Espresso, Chromium, Gecko, Safari, and Mac2 are desirable but are not implemented selection targets yet. Add a driver profile, symptom-specific references, verification steps, and completion evidence before advertising any of these drivers in the troubleshooting Skill description.
- See AGENTS.md for repository-level execution rules and skill entrypoint guidance.
- Use the selected
SKILL.mdand its routedcontexts/assets for target-specific step order. - Renma metadata and
renma.config.jsonare retained for optional maintainer analysis; Renma is not required to run or complete any repository skill.