A
licenseB
qualityA
maintenanceA Model Context Protocol (MCP) server and CLI that provides tools for agent use when working on iOS and macOS projects.
83
2,581 npm
6,536
MIT
adb-mcp
by iksnerd
Related Servers
Alternatives to adb-mcp
- AlicenseAqualityNot gradedmaintenanceEnables browser automation through Playwright using accessibility tree snapshots instead of screenshots. Supports web scraping, form interactions, testing, and connecting to existing browser sessions with logged-in accounts.238,020 npm5-
Related Servers
- AlicenseAqualityCmaintenanceEnables coding agents to build, drive, and debug Android apps on a real device, with screenshots, UI dumps, local vision model analysis, and lock-safe Gradle builds.21MIT
- AlicenseBqualityBmaintenanceEnables an AI agent to drive an Android app under development on a local emulator or connected device through adb, observing the screen and performing taps, swipes, text entry, key presses, waits, screenshots, logcat and crash inspection. It is restricted to debug builds installed via Run in Android Studio, so agents can exercise only the app being developed.2313 npmMIT
- AlicenseAqualityDmaintenanceEnables an agent to inspect and interact with Android emulators or physical devices via ADB, capturing UI snapshots, tapping nodes, typing text, and reading app logs.10MIT
- AlicenseAqualityDmaintenanceEnables AI agents to build, deploy, drive, and debug Android apps — managing Gradle builds, emulators, adb deployment, logcat capture, and full UI automation.5373 npmMIT
- AlicenseBqualityAmaintenanceEnables AI agents to automate Android and iOS devices, including emulators and physical devices, by capturing screenshots, reading UI accessibility trees, performing touch and text input, and managing app lifecycle and logs.183MIT
- AlicenseNot gradedqualityBmaintenanceEnables coding agents to build, install, and launch Android apps on emulators and physical devices, automate UI interactions, inspect logcat/crash diagnostics, measure JaCoCo coverage, and discover or scaffold projects through an MCP server.6,270 npmMIT
TDQS
A3.7/5.0
Scored across 80 tools
Disambiguation3/5
Descriptions go out of their way to cross-reference siblings (describe_ui vs render_stats, screenshot vs describe_ui, reload_app vs open_dev_menu, swipe vs drag), which rescues a lot. But there are several genuinely confusable families: tap / tap_on_text / tap_element (three ways to press the same pixel), logcat / start_logcat_capture / stop_logcat_capture / clear_logcat, wait / wait_for_text / wait_for_boot, gradle_build / build_and_run / build_status, and get_coverage_report vs get_file_coverage. An agent must read long prose to pick correctly, and with 80 tools the wrong pick is easy.
Naming Consistency4/5
Everything is snake_case, which is the dominant and predictable convention here. There is minor drift in verb placement (verb_noun like press_key, send_sms, launch_app vs noun_verb like phone_call, fingerprint_touch, avd_snapshot) and a handful of bare nouns (doctor, screenshot, logcat, wait, swipe, tap). Still readable and mostly memorable.
Tool Count2/5
80 tools is far past what an agent can hold in context, even granted that the surface spans device driving, emulator emulation, Gradle build/test/coverage, and session defaults. Several clusters could be collapsed (three session_* tools, four logcat tools, three tap variants, build_status folded into gradle_build, get_file_coverage into get_coverage_report). The breadth is real but the granularity is excessive.
Completeness4/5
Coverage of the Android-automation domain is unusually thorough: device/app lifecycle, permissions, logs, sensors, radio/battery/location, biometrics and lock screens, emulator snapshots, plus Gradle build/variant/test/coverage. Minor gaps: no multi-touch pinch/zoom gesture, no 'scroll until element found' helper, no device file listing (only push/pull), no notification-shade interaction, and the accessibility-bridge install referenced by via_accessibility has no tool of its own.
Maintenance
ActivityMaintained
ResponsivenessResponsive