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 commands/kaeawc/auto-mobile/manual-testgit clone --depth 1 https://github.com/kaeawc/auto-mobileWrote 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/commands/kaeawc/auto-mobile/manual-test)<a href="https://agentmods.dev/commands/kaeawc/auto-mobile/manual-test"><img src="https://agentmods.dev/badge/commands/kaeawc/auto-mobile/manual-test.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 | $0.00000 | $0.00725 |
| Opus 5 | $0.00000 | $0.00362 |
| Sonnet 5 | $0.00000 | $0.00145 |
| Haiku 4.5 | $0.00000 | $0.00072 |
Grade A, and why
manual-test 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 — 51 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run one AutoMobile manual-test iteration. Load and follow the
manual-test skill (skills/manual-test/SKILL.md) — it is the
source of truth for the procedure and the hard-won gotchas; this command only
sets up the inputs.
Start point
$ARGUMENTS is the starting point to test forward from — a commit SHA, a git
tag/milestone, or a date. If it is empty, ask the user for one before doing
anything else (offer the latest release tag as the default:
git tag | sort -V | tail -3). Resolve it to a git ref and a date, then compute
the range <START>..origin/main.
What to do
Execute the skill's phases in order:
- Scope — enumerate merged PRs and closed issues since
<START>, classify each as bug-fix (reproduce → confirm fixed) or feature/spec (exercise → confirm delivered), and map the changedsrc/surface to affected MCP tools. Present the checklist. - Rebuild ALL necessary components — rebase + fast-forward the daemon's main
checkout;
bun run build+ regenerate schemas; rebuild the Android ctrlproxy APK ifandroid/control-proxy/**changed and the iOS runner ifios/control-proxy/**changed; rebuild the playground SDK app only if testing SDK features (standard gradle output, not the grit/gojvm variants). Never trust the daemon version string — verify by build hash / dist mtime. - Restart the daemon with exactly the flags this run needs (
--embedded-sdk,--network-mockable, APK/iOS runner overrides). Kill stray daemons first; if a competing worktree daemon keeps respawning, mark SDK-gated tests blocked. If the MCP tools are build-skewed after the restart, drive tools via the CLI (bun dist/src/index.js --cli <tool> --<param> <value>). - Exercise tool calls — Android first, then iOS, one device at a time. For each
item, ground the result in an observed field or adb/simctl ground truth, not the
tool's
successflag. Sweep the changed surface for regressions. - Report & file — per-item FIXED/PASS/NOT-FIXED/REGRESSED/BLOCKED table with evidence; file issues for regressions/unfixed (repro + root cause + suggested fix); comment verification results on the closed issues / merged PRs.
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 · 51 lines · 0 tokens per session scan A a1b0d3372cd6
manual-test is a command published in the GitHub repository kaeawc/auto-mobile (45 stars, last pushed yesterday), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 725 tokens. 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 commands, from other repositories
test-simulator
Launch simulator testing agent for test scenarios and visual verification.
run-tests
Run XCUITests and parse results using the test-runner agent.
flutter-test
Run Flutter/Dart tests, report failures, and incrementally fix test issues. Covers unit, widget, golden, and integration tests.
qa-ios-screenshot
Take a simulator screenshot and analyze it for visual issues.
test-feature
Test a React Native feature on the running simulator/emulator. Verifies UI, user flows, and internal state. Generates a persistent Maestro test file.
ui-test
Run visual UI tests by comparing device screenshots against benchmark images. Supports ADB actions (tap, type, swipe, back, wait, launch) to navigate through app flows.