agentleFS
molt-botmolt-bot/clawdbot/.agents/skills/ios-e2e/SKILL.md

ios-e2e

Skill395k starsChanged todayMITView on GitHub

iOS app on the simulator - build, run, pair with an isolated Gateway, drive the UI, screenshots, logs and crashes, focused tests, and end-to-end proof.

  • Installs packages

3 other projects use a version of this.

Adopted by

Steady: 0 new projects took it in the last 4 weeks, none the 4 weeks before.thin data

Week of Sep 20: 0 new projectsWeek of Sep 27: 0 new projectsWeek of Oct 4: 0 new projectsWeek of Oct 11: 0 new projectsWeek of Oct 18: 0 new projectsWeek of Oct 25: 0 new projectsWeek of Nov 1: 0 new projectsWeek of Nov 8: 0 new projects
Sep 2026Oct 2026Nov 2026
New projects taking a version of this file each week, over the 8 weeks our latest run read: only 0 in all, so this is a hint, not a trend.

Most recently taken up by

Who's talking about molt-bot/clawdbot

The biggest 5 of 18 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 18

Talked about on the web (18 times).

On the web · 18

13 more on the web

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Run the touched suites: pnpm ios:test <Suite> (selectors and IOS_SIM: apps/ios/README.md).

    line 55
  2. Very specific:

    Start the isolated Gateway (one-time profile in Setup) as a long-running service: node scripts/run-node.mjs --profile ios-sim gateway …

    line 62
  3. Distinctive:

    Stop the Gateway and delete the simulators.

    line 66
  4. Very specific:

    Check: mobilebuildmcp simulator list-schemes lists OpenClaw, and mobilebuildmcp simulator build succeeds.

    line 94

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “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

iOS E2E

Work from the checkout root. A local .mobilebuildmcp/config.yaml (Setup) supplies the project, scheme and DerivedData, so mobilebuildmcp commands need no project flags.

Two values recur:

  • $sim: the simulator UDID (pnpm ios:sim:pair prints one). Pass --simulator-id "$sim" to every mobilebuildmcp call; without it the config default targets the simulator named iPhone 17.
  • $bundle: the bundle ID, which depends on the signing team. Read it from the build: bundle=$(/usr/libexec/PlistBuddy -c 'Print :CFBundleIdentifier' apps/ios/build/DerivedData/Build/Products/Debug-iphonesimulator/OpenClaw.app/Info.plist).

Build and run

  1. Build: mobilebuildmcp simulator build. After editing apps/ios/project.yml, run pnpm ios:gen first.
  2. Install and launch with simctl, which takes launch arguments (pnpm ios:run takes none):

    xcrun simctl install "$sim" apps/ios/build/DerivedData/Build/Products/Debug-iphonesimulator/OpenClaw.app
    xcrun simctl launch --terminate-running-process "$sim" "$bundle" --openclaw-reset-onboarding

    mobilebuildmcp simulator build-and-run and launch-app attach a log capture that never exits.

  3. Debug launch arguments: grep -rhoE '"--openclaw-[a-z-]+"' apps/ios/Sources | sort -u. --openclaw-screenshot-mode plus an --openclaw-*-fixture flag shows a fixture screen without a Gateway; -AppleLanguages "(de)" -AppleLocale de_DE runs another language. The Debug URL scheme is openclaw-debug://.

Drive the UI

  1. mobilebuildmcp ui-automation snapshot-ui --style minimal lists tappable targets as ref|action|role|label|value|id. Text that is not a control (chat bubbles, row titles) appears only with --output json.
  2. Act on a ref with tap, type-text, long-press (--duration is in milliseconds), swipe or gesture. Wait with wait-for-ui: --predicate exists --label '…' --role button for an element, --predicate textContains --text '…' for text. A wait selector that matches several elements fails; add --role or --identifier. Keep --timeout-ms under 30000: the daemon ends any request after 30 s with DAEMON_TRANSPORT_FAILED; repeat the wait for longer.
  3. Refs belong to one snapshot; take a fresh one after every screen change.
  4. See the screen: mobilebuildmcp simulator screenshot, then read the printed JPEG.

Gotchas:

  • Onboarding and sheets also list the screens behind them; match by label or id and take the last match.
  • On iPhone the sidebar overlays the content: tap RootTabs.Sidebar.Show before its rows.
  • xcrun simctl openurl queues a system "Open in OpenClaw?" alert per call; tap Open or Cancel for each.
  • System UI outside the app's process (photo picker, share sheet) is missing from snapshots. Tap it by coordinates with the AXe binary bundled in MobileBuildMCP: "$(brew --prefix mobilebuildmcp)/libexec/bundled/axe" tap -x <x> -y <y> --udid "$sim". Coordinates are points: screenshot pixels × screen width in points ÷ screenshot width.

Logs and crashes

Act first, then read a bounded window:

xcrun simctl spawn "$sim" log show --last 2m --style compact --predicate 'subsystem BEGINSWITH "ai.openclaw"'

Crashes land in ~/Library/Logs/DiagnosticReports/OpenClaw-*.ips; the first OpenClaw frame of the triggered thread names the culprit. App preferences and databases live in the app container (xcrun simctl get_app_container "$sim" "$bundle" data); simctl spawn … defaults read cannot see them. LLDB and xctrace: DEBUG.md.

Tests

  1. Run the touched suites: pnpm ios:test <Suite> (selectors and IOS_SIM: apps/ios/README.md).
  2. Prove a regression test red on the base: commit the change, check out the base version of the changed source files (git checkout <base> -- <paths>), run the suite, then git checkout HEAD -- <paths>.

Done: the touched suites pass, and each new regression test fails on the base.

Prove a change end to end

  1. Start the isolated Gateway (one-time profile in Setup) as a long-running service: node scripts/run-node.mjs --profile ios-sim gateway --bind loopback. It never touches the operator's Gateway or ~/.openclaw. While it runs, call the CLI as node openclaw.mjs --profile ios-sim …; run-node refuses to rebuild under a running Gateway.
  2. Pair a fresh simulator through the real onboarding UI: sim=$(pnpm --silent ios:sim:pair | tail -1) (--app for another build, --profile for another profile).
  3. Exercise the change through visible controls. A chat round trip: type into chat-message-input, tap chat-send-message, then wait for a word only the reply contains.
  4. Compare builds: the same steps on main and on the change, each on its own fresh simulator, with matching screenshots.
  5. Stop the Gateway and delete the simulators.

Node commands, permissions, auth failures, outages, rejected sends, upgrades and iPad layout: PROOF.md, which also lists what only a device proves. Chat scrolling, streaming, keyboard and relaunch stability (rows jumping, drifting or rebuilding), measured against a scripted model: BEHAVIOR.md.

Done: screenshots or recorded output show the behavior on the change and its absence on main; the Gateway port is closed and no openclaw-pair-* simulator remains.

Setup

One-time per machine; each step ends on a check.

  1. Xcode, simulator runtimes, xcodegen and the Watch Rust toolchain: apps/ios/README.md prerequisites. Check: xcrun simctl list devices available shows iPhone 17.
  2. SwiftFormat and SwiftLint at the repo pin: scripts/install-swift-tools.sh <dir on PATH>. Check: scripts/check-swift-tools.sh exits 0.
  3. MobileBuildMCP, Sentry's XcodeBuildMCP CLI: brew tap getsentry/xcodebuildmcp && brew install getsentry/xcodebuildmcp/mobilebuildmcp. These steps were written against 2.7.1. Check: mobilebuildmcp --version.
  4. Checkout: pnpm install --frozen-lockfile, then pnpm ios:gen (the Xcode project is generated, not committed). Then create .mobilebuildmcp/config.yaml and add /.mobilebuildmcp/ to .git/info/exclude:

    schemaVersion: 1
    enabledWorkflows: [simulator, simulator-management, ui-automation, debugging]
    sentryDisabled: true
    sessionDefaults:
      projectPath: ./apps/ios/OpenClaw.xcodeproj
      scheme: OpenClaw
      configuration: Debug
      platform: iOS
      simulatorName: iPhone 17
      derivedDataPath: ./apps/ios/build/DerivedData # shared with pnpm ios:run

    Check: mobilebuildmcp simulator list-schemes lists OpenClaw, and mobilebuildmcp simulator build succeeds.

  5. The ios-sim Gateway profile, with the Gateway stopped (oc = node openclaw.mjs --profile ios-sim): oc config set gateway.mode local, oc config set gateway.port 19031, oc config set gateway.auth.mode token, oc config set gateway.auth.token "$(openssl rand -hex 24)". Chat replies need a model for its main agent: oc models auth login or add with test credentials. Check: the Gateway starts, and pnpm ios:sim:pair ends on the chat screen.

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.