Dev & Engineering mobile-testingexploratory-testingreact-nativeios-testingandroid-testingagent-devicebug-reporting

Mobile App Dogfood Tester

Systematically explore iOS and Android apps and produce bug reports backed by screenshots and reproducible steps.

FollowSkills review · FSRS-2.0
Not recommended
46/ 100 5-point scale 2.3 / 5
Trust10 / 25 · 2.0/5

The skill scopes its declared tool access to agent-device commands and requires the user to provide OTP/email codes, showing some permission and confirmation awareness. However, it can launch apps, authenticate, execute create/edit/delete flows, inspect logs, and write screenshots, videos, and reports without specifying sensitive-data handling, external data flows, rollback, or cleanup, so points are deducted.

Reliability8 / 20 · 2.0/5

The workflow, commands, and report template are broadly consistent and include resnapshotting, targeted log inspection, and abnormal-state exploration. However, agent-device version/install validation is absent, and failures involving unavailable devices, authentication, missing paths, or missing templates lack diagnosable recovery guidance; the static ceiling therefore applies and points are deducted.

Adaptability10 / 15 · 3.3/5

Triggers, platforms, required target app, authentication, scope, and output are reasonably clear for mobile exploratory testing on iOS and Android. Non-fit cases, platform-specific boundaries, Chinese-language interaction support, mainland-China network reachability, and environmental dependency constraints are not documented, so points are deducted.

Convention8 / 15 · 2.7/5

The skill uses useful progressive sections and supplies an issue taxonomy plus a report template. But it lacks skill-local versioning, changelog, explicit maintenance ownership, dependency installation instructions, FAQs, and known limitations; placeholder and path assumptions are also unexplained, so points are deducted.

Effectiveness6 / 15 · 2.0/5

The skill offers an actionable mobile exploration process, reproducibility guidance, and a structured report that could be directly useful. Static evidence contains no real execution results or representative completed report, while agent-device availability and substantial human judgment remain prerequisites, so correctness, completeness, and marginal benefit are not verifiable and points are deducted.

Verifiability4 / 10 · 2.0/5

The skill, taxonomy, and template are auditable and provide limited internal corroboration. There is no key-path test suite, CI execution evidence, completed example report, or third-party reproduction record, so evidence strength is limited and points are deducted.

Evidence confidence:Low Reviewed Jul 20, 2026 Reviewed revision f6b87e35a35c
Before you use it
  • Execution may mutate the target app through create, edit, delete, submit, and login flows; the skill does not require pre-action confirmation, isolated test accounts, or data recovery planning.
  • Screenshots, videos, and logs may contain personal, credential, or business data; no redaction, retention, or access-control guidance is provided.
  • agent-device installation, version, device connectivity, and network prerequisites are not validated within the skill, and failure feedback and recovery steps are limited.
  • No runtime evidence is supplied beyond the static files, so report quality, issue-detection rate, and iOS/Android parity require human validation.
Review evidence [1][2][3][4][5][6]
See the full review method →

What does this skill do, and when should you use it?

Dogfood is an exploratory testing skill for mobile apps on iOS and Android. It uses agent-device to launch an app, map navigation, exercise core flows, check edge states, and document observed problems. Findings can include step screenshots and optional reproduction videos. It fits teams that need runtime QA evidence, provided agent-device and a runnable target app are available.

It takes a target app as required input and infers or asks for the platform. It uses agent-device to create sessions, open the app, capture snapshots and screenshots, perform interactions such as filling fields and pressing controls, wait for state changes, inspect app state and logs, and record evidence when an issue is found. The workflow ends with a structured report containing issue totals, severity counts, highest-risk findings, and reproduction evidence.

  1. A mobile developer checks primary navigation and core flows before an iOS or Android release.
  2. A QA engineer explores end-to-end journeys such as creating, editing, submitting, deleting, or recovering data.
  3. A product team validates empty, loading, error, offline, and denied-permission states.
  4. A team needs reproducible runtime bug reports with screenshots, numbered steps, and optional video.

What are this skill's strengths and limitations?

Pros
  • Provides a systematic workflow covering launch, navigation, core journeys, and edge states.
  • Preserves screenshots, detailed steps, and optional reproduction video for each finding.
  • Supports both iOS and Android and includes session, logging, and report wrap-up guidance.
Limitations
  • Requires agent-device plus a runnable mobile app and device environment.
  • The SKILL.md uses Bash(agent-device:*) and Bash(npx agent-device:*) permission syntax, so other clients may need adaptation.
  • The source does not establish compatibility across specific iOS or Android versions, device models, or app types.
  • It explicitly avoids reading the app's source code, so it cannot add static-analysis evidence.

How do you install this skill?

The README says the repository can provide bundled Codex plugins through .agents/plugins/marketplace.json and can also be cloned for use with its skill files. This skill is located at plugins/vendored/.agents/skills/dogfood/SKILL.md; the source does not document a dedicated dogfood installation command. agent-device must be available in the execution environment.

How do you use this skill?

Example trigger: "Dogfood Settings on iOS and focus on onboarding". A target app is required; specify ios or android when the platform cannot be inferred. You may also provide a session name, output directory, scope, and authentication details, for example: "Test com.example.app on Android, focus on profile, sign in with [email protected]". If an OTP or email code is required, the workflow asks the user for it.

FAQ

What is this skill best suited for?
It is suited to exploratory QA, bug hunts, and end-to-end runtime validation of iOS or Android apps, not general code review or static analysis.
What permissions or dependencies are required?
It requires shell execution, filesystem access for artifacts, and agent-device; the SKILL.md also permits invoking agent-device through npx.
What happens if the app or platform is missing?
The target app is mandatory. If the platform cannot be inferred, the user must provide ios or android.
What evidence does it produce?
It produces a structured report and screenshots; interactive findings may also include reproduction videos and detailed steps.

More skills from this repository

All from callstackincubator/agent-skills

Dev & Engineering

React Native Migration Assessor

Evidence-led decisions for React Native migration readiness

Dev & Engineering

React Native Brownfield Migration

Guides incremental React Native or Expo adoption inside existing native iOS and Android apps.

Automation & Ops

React Native GitHub Actions Artifacts

Build React Native simulator artifacts in the cloud and retrieve them from GitHub.

Dev & Engineering

React Native TV Review Guide

Systematically review TV-specific focus, playback, performance, and accessibility issues in React Native apps.

Dev & Engineering

React Native Performance Playbook

Actionable rules for building faster, more reliable React Native and Expo apps.

Dev & Engineering

React Native Performance Guide

Diagnose and improve React Native performance with measured, evidence-based guidance.

Dev & Engineering

React Native Upgrade Assistant

A guided workflow for upgrading React Native apps through template diffs, dependency migration, and native configuration changes.

Dev & Engineering

React Navigation UI Guide

Practical React Navigation 7 patterns for building and configuring React Native navigation interfaces.

Dev & Engineering

React Native Library Builder

Scaffold publishable React Native libraries or local native modules with a guided workflow.

Dev & Engineering

Cross-Platform Device Operator

Enables AI coding assistants to inspect and operate Apple-platform and Android apps.

Dev & Engineering

React Native Test Guide

Helps AI write dependable React Native component tests

Dev & Engineering

Skill Spec Validator

Checks Agent Skills against the agentskills.io spec and Claude Code best practices.

Dev & Engineering

GitHub PR Workflow Assistant

Manage pull requests, stacked branches, and merges through the gh CLI.

Related skills