agentleFS
Sign inStart free
Simple-With-UsSimple-With-Us/BotFleet/skills/phone-harness/SKILL.md

phone-harness

Skill4 starsChanged 10 days agoApache-2.0View on GitHub

Control, inspect, test, or automate a physical Android phone connected to BotFleet over authorized USB debugging. Use for explicit Android, USB phone, ADB, mobile-app, tapping, typing, swiping, scrolling, screenshot, or phone-screen requests.

6 other projects use a version of this, 4 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. 1 is common, 0 are distinctive, and 6 are very specific to this project.

  • 1

    Common

    Most of the 5 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use the phone tools for every requested Android action.

    line 8
  2. Very specific:

    Call status before the first action.

    line 11
  3. Very specific:

    Call open_app with the human app name.

    line 13
  4. Very specific:

    Use list_apps only when the name is ambiguous.

    line 14
  5. Very specific:

    Call read_screen before choosing a target and after every action.

    line 15
  6. Very specific:

    Use swipe, type_text, and press for interaction, verifying each step.

    line 18
  7. Common:

    Never enter passwords, payment details, government identifiers, or one-time codes.

    line 25

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Phone Harness

Use the phone tools for every requested Android action. Never replace them with Bash, raw adb, subprocess, an emulator, network ADB, or Tailscale.

  1. Call status before the first action. If the device is missing or unauthorized, stop and relay the physical setup instruction.
  2. Call open_app with the human app name. Do not scan the launcher first. Use list_apps only when the name is ambiguous.
  3. Call read_screen before choosing a target and after every action. Prefer tap_text; use screenshot and pixel tap only when accessibility text cannot identify the target.
  4. Use swipe, type_text, and press for interaction, verifying each step.

The tools operate the user's real phone. Navigate and read only what the task needs. Stop before sending, posting, purchasing, booking, deleting, changing settings, entering protected information, or accepting an unexpected legal or security prompt unless the user has explicitly authorized that exact action.

Never enter passwords, payment details, government identifiers, or one-time codes. Ask the user to complete protected-input steps directly on the phone.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free