android-claude-code-skills

v1.7.0

Published

a month ago

Reusable Android development skills for Claude Code, OpenAI Codex, Gemini CLI, and Cursor — translation sync, code review, APK analyzer

Android Claude Code Skills

Languages: English | 中文 | 日本語

A repository of reusable Android development skills, compatible with Claude Code, OpenAI Codex, Gemini CLI, and Cursor.

Repository Information

GitHub: https://github.com/aihip/android-claude-code-skills
Author: aihip
License: MIT
Current Version: 1.7.0
Changelog: CHANGELOG.md

OpenAI Codex Compatibility

This project primarily provides Android development skills, and the skill files are maintained in a format compatible with both OpenAI Codex and Claude Code:

skills/<skill-name>/SKILL.md includes YAML frontmatter with name and description
Optional skills/<skill-name>/agents/openai.yaml provides Codex UI metadata
Existing Claude plugin structure (.claude-plugin/) is preserved for dual compatibility

Installation

# Add this repository as a marketplace
/plugin marketplace add aihip/android-claude-code-skills

# Install the plugin
/plugin install android-claude-code-skills@android-claude-code-skills

Updating Plugin

IMPORTANT: Only this method works for updating:

# Step 1: Update marketplace
/plugin marketplace update aihip/android-claude-code-skills

# Step 2: Update plugin
/plugin update android-claude-code-skills

Note: Other update methods may not fetch the latest version. Always use marketplace update first.

What Gets Updated

When you update the plugin, you'll receive:

New skills added to the repository
Improvements to existing skills
Bug fixes and enhancements

Verify Update Success

After updating, check the version:

# View installed plugins and versions
/plugin list

# You should see:
# android-claude-code-skills  v1.7.0

Compare with the latest version on GitHub: https://github.com/aihip/android-claude-code-skills/blob/main/.claude-plugin/plugin.json

Available Skills

Android Project Analyzer

Analyze an Android project codebase and produce a comprehensive technical report covering structure, features, architecture, dependencies, code quality, and potential issues.

Usage:

Analyze the Android project in the current directory.

Features:

Project overview: applicationId, SDK versions, module structure, file counts
Architecture detection: MVVM / MVI / Clean Architecture, DI framework, navigation, Compose vs View
Feature inventory: entry points, packages, deep links
Full dependency list with versions grouped by category (UI / Network / DI / Database / Testing)
Code quality metrics: test coverage surface, TODO/FIXME count, force-unwrap (!!) count, large files
Potential issues: hardcoded secrets, cleartext HTTP, memory leaks, crash risks, outdated libraries
Prioritized recommendations

Trigger phrases:

analyze android project
analyze this project
project architecture review
dependency audit
code quality review
project health check
分析安卓项目
项目架构分析
依赖审计
代码质量分析

Android Translation Sync

Synchronize Android project multilingual resources from Excel spreadsheet.

Usage:

Please sync translations from excel: /path/to/translations.xlsx

Features:

Read Excel file with translations (English + other languages)
Generate unique keys (10-20 chars) based on English text
Avoid conflicts with existing keys in strings.xml
Update all language files (values, values-zh, values-es, etc.)
New entries appended at bottom, existing keys updated

Example:

你：请帮我同步多语言翻译，Excel 文件是 ./translations/strings.xlsx

Claude：好的，我来处理...
- Reading Excel file...
- Checking existing strings.xml...
- Generating new keys...
- Updating all language files...

Trigger phrases:

sync translations from excel
update android strings from excel
多语言翻译同步
更新strings.xml

Android Change Review

Review Android code changes after coding to catch crash risks, boundary-condition bugs, and regressions.

Usage (staged changes):

git add .

Then ask:

Review my staged changes for crash risks and boundary conditions.

Usage (specific commit):

Please review commit abc1234 for crash risks and regressions.

Features:

Review staged changes (git diff --cached) after git add .
Review a specific commit (git show <commit-id>)
Focus on crash paths, boundary conditions, and regressions
Android-focused checks (lifecycle, nullability, threading, permission gating)
Findings-first output with file/line references

Trigger phrases:

review staged changes
review commit-id
code review after coding
检查当前修改代码
边界条件检查
崩溃风险检查

APK Analyzer

Parse and audit Android APK files: extract metadata, classify permissions by risk, verify signatures, inspect exported components, detect native libraries and third-party SDKs, and produce a security report.

Usage:

Analyze this APK: /path/to/your.apk

Features:

Extract package name, version code/name, min/target SDK from APK
Classify all permissions as Dangerous / High-Risk / Normal
Verify V1/V2/V3 signature schemes and print certificate details
Detect debug keystore by SHA-1 fingerprint
Identify exported Activities, Services, Receivers, and Providers
Check security flags: debuggable, allowBackup, cleartext traffic
Detect native library ABIs and third-party SDKs (Flutter, React Native, Firebase, etc.)
Includes a full shell script (apk-analyze.sh) and Python (androguard) analysis path
Produces a structured security audit report with CRITICAL / HIGH / MEDIUM / INFO findings

Trigger phrases:

analyze apk
check apk permissions
verify apk signature
audit apk security
apk信息提取
apk权限分析
apk签名检查
apk安全审计

Third-Party Skills

Community-maintained plugins that extend Claude Code with additional workflows.

review-loop — Automated Code Review Loop

Source: hamelsmu/claude-review-loop

A Claude Code plugin that adds an automated, two-phase code review loop to every task. After Claude finishes implementing a task, a stop hook automatically triggers an independent Codex review, then asks Claude to address the findings — giving every change a second opinion before you accept it.

How it works:

Task phase — You describe a task, Claude implements it.
Review phase — When Claude finishes, the stop hook runs Codex (codex exec) for an independent review, writes findings to reviews/review-<id>.md, then asks Claude to address the feedback.
Claude resolves the items it agrees with and exits cleanly.

State is tracked in .claude/review-loop.local.md (add to .gitignore).

Review coverage:

| Area | What gets checked | |---|---| | Code quality | Organization, modularity, DRY, naming | | Test coverage | New tests, edge cases, test quality | | Security | Input validation, injection, secrets, OWASP Top 10 | | Docs & agent harness | AGENTS.md, CLAUDE.md symlinks, telemetry, type system | | UX & design | E2E tests, visual quality, accessibility (UI projects) |

Requirements:

Claude Code CLI
jq — brew install jq (macOS) / apt install jq (Linux)
Codex CLI (recommended, falls back to Claude self-review if absent) — npm install -g @openai/codex

Installation:

# From within a Claude Code session
/plugin marketplace add hamelsmu/claude-review-loop
/plugin install review-loop@hamel-review

Or from the CLI directly:

claude plugin marketplace add hamelsmu/claude-review-loop
claude plugin install review-loop@hamel-review

Usage:

# Start a review loop for a task
/review-loop Add user authentication with JWT tokens and test coverage

# Cancel an in-progress review loop
/cancel-review

Configuration:

| Environment Variable | Default | Description | |---|---|---| | REVIEW_LOOP_CODEX_FLAGS | --dangerously-bypass-approvals-and-sandbox | Flags passed to codex. Use --sandbox workspace-write for safer sandboxed reviews. |

The stop hook timeout is 900 seconds (15 min) by default — adjust in hooks/hooks.json if your reviews take longer.

Logs: Execution logs with timestamps, codex exit codes, and elapsed times are written to .claude/review-loop.log (gitignored).

claude-codex — Multi-AI Orchestration Pipeline

Source: Z-M-Huang/claude-codex Note: This project has moved to Z-M-Huang/vcp. Future development continues there.

A Claude Code plugin providing a multi-AI orchestration pipeline that runs your code through three independent reviewers — Claude Sonnet, Claude Opus, and Codex — before accepting any change. Based on the professional principle that no code should be deployed after only one reviewer.

Why multi-AI review?

| Reviewer | What It Catches | |---|---| | Claude Sonnet | Obvious bugs, security basics, code style | | Claude Opus | Architectural issues, subtle bugs, edge cases | | Codex | Fresh perspective from a different AI model |

Each reviewer checks for OWASP Top 10 vulnerabilities, proper error handling, and code quality. The loop-until-approved model means code doesn't proceed until all three reviewers give the green light.

Available skills:

| Skill | Purpose | |---|---| | multi-ai | Full feature development pipeline (requirements → plan → implement → review) | | bug-fix | Bug-fix pipeline with dual root-cause analysis + Codex validation |

Custom agents used internally:

| Agent | Model | Role | |---|---|---| | requirements-gatherer | Opus | Business Analyst + PM hybrid | | planner | Opus | Architect + Fullstack hybrid | | plan-reviewer | Sonnet + Opus | Architecture, Security & QA validation | | implementer | Sonnet | Fullstack + TDD + quality implementation | | code-reviewer | Sonnet + Opus | Security, Performance & QA validation | | root-cause-analyst | Sonnet + Opus | Parallel bug diagnosis (bug-fix pipeline) |

Requirements:

Claude Code CLI
Codex CLI — npm install -g @openai/codex
Bun (used for cross-platform JSON processing, replaces jq)

Installation:

# Step 1: Add marketplace
/plugin marketplace add Z-M-Huang/claude-codex

# Step 2: Install plugin (user scope — available in all projects, recommended)
/plugin install claude-codex@claude-codex --scope user

# Step 3: Add .task to .gitignore
echo ".task" >> .gitignore

Usage:

# Feature development pipeline
/claude-codex:multi-ai Add user authentication with JWT tokens

# Bug-fix pipeline
/claude-codex:bug-fix Login fails silently when session token expires

Always use the full namespace claude-codex:<skill> when invoking from external projects. Or describe the task naturally and Claude will invoke the right skill.

Pipeline flow (/multi-ai):

Requirements — Specialist agents explore in parallel; requirements-gatherer synthesizes
Planning — planner agent creates implementation plan
Plan Reviews — plan-reviewer (Sonnet + Opus) + Codex gate
Implementation — implementer iterates until tests pass
Code Reviews — code-reviewer (Sonnet + Opus) + Codex final gate
Complete — Reports results

Pipeline flow (/bug-fix):

Dual RCA — Two root-cause-analyst agents (Sonnet + Opus) analyze in parallel
Consolidation — Orchestrator synthesizes both analyses into a fix plan
Codex Validation — Codex reviews the consolidated RCA and fix plan
Implementation — Minimal fix targeting the root cause
Code Reviews — code-reviewer (Sonnet + Opus) + Codex gate

Pipeline enforcement (task dependencies):

1. Implement  →  2. Review (Sonnet)  →  3. Review (Opus)  →  4. Review (Codex)
                      ↓ needs_changes?
                 Create fix task → same reviewer re-validates → continue

Default limits:

| Setting | Default | |---|---| | Plan review loop limit | 10 iterations | | Code review loop limit | 15 iterations | | Auto-resolve attempts | 3 retries |

License: GPL-3.0 with attribution requirement (author: Z-M-Huang).

Using Skills in Codex CLI

The skills in this repository are natively compatible with OpenAI Codex CLI — the SKILL.md files already include the required YAML frontmatter (name, description) and each skill has an agents/openai.yaml for UI metadata.

Installation

Codex discovers skills from .agents/skills/ directories. Copy the skills to either user-level or project-level:

# User-level — available in ALL your projects (recommended)
mkdir -p ~/.agents/skills
cp -r skills/android-project-analyzer ~/.agents/skills/
cp -r skills/android-translation-sync ~/.agents/skills/
cp -r skills/android-change-review ~/.agents/skills/
cp -r skills/apk-analyzer ~/.agents/skills/

# Project-level — this project only
mkdir -p .agents/skills
cp -r skills/android-project-analyzer .agents/skills/
cp -r skills/android-translation-sync .agents/skills/
cp -r skills/android-change-review .agents/skills/
cp -r skills/apk-analyzer .agents/skills/

Or use symlinks to always stay in sync with repo updates:

ln -s "$(pwd)/skills/android-project-analyzer" ~/.agents/skills/
ln -s "$(pwd)/skills/android-translation-sync" ~/.agents/skills/
ln -s "$(pwd)/skills/android-change-review" ~/.agents/skills/
ln -s "$(pwd)/skills/apk-analyzer" ~/.agents/skills/

Usage

Explicit invocation (type $ to open the skill selector):

$android-project-analyzer  analyze the current project

$android-translation-sync  ./translations/strings.xlsx

$android-change-review  review my staged changes

$apk-analyzer  ./build/outputs/apk/release/app-release.apk

Implicit invocation — Codex auto-selects the skill based on your description (allow_implicit_invocation: true):

Analyze the Android project architecture and dependencies.

Sync translations from Excel: ./translations/strings.xlsx

Review my staged changes for Android crash risks and boundary conditions.

Analyze this APK for permissions and security issues: ./app-release.apk

Verify Installation

# In a Codex CLI session
/skills
# Should list: android-project-analyzer, android-translation-sync, android-change-review, apk-analyzer

Using Skills in Cursor

The skills in this repository are natively designed for Claude Code CLI. For Cursor users, pre-converted .mdc rule files are provided in the cursor-rules/ directory.

Available Cursor Rules

| File | Skill | |---|---| | cursor-rules/android-project-analyzer.mdc | Android Project Analyzer | | cursor-rules/android-translation-sync.mdc | Android Translation Sync | | cursor-rules/android-change-review.mdc | Android Change Review | | cursor-rules/apk-analyzer.mdc | APK Analyzer |

Installation

Copy the .mdc files into your Android project's .cursor/rules/ directory:

# In your Android project root
mkdir -p .cursor/rules

curl -o .cursor/rules/android-project-analyzer.mdc \
  https://raw.githubusercontent.com/aihip/android-claude-code-skills/main/cursor-rules/android-project-analyzer.mdc

curl -o .cursor/rules/android-translation-sync.mdc \
  https://raw.githubusercontent.com/aihip/android-claude-code-skills/main/cursor-rules/android-translation-sync.mdc

curl -o .cursor/rules/android-change-review.mdc \
  https://raw.githubusercontent.com/aihip/android-claude-code-skills/main/cursor-rules/android-change-review.mdc

curl -o .cursor/rules/apk-analyzer.mdc \
  https://raw.githubusercontent.com/aihip/android-claude-code-skills/main/cursor-rules/apk-analyzer.mdc

Or clone this repo and copy manually:

cp android-claude-code-skills/cursor-rules/*.mdc your-android-project/.cursor/rules/

Usage in Cursor

All rules are Agent-requested type — Cursor's AI automatically activates them when it detects a matching request. You can also trigger them explicitly:

# Project analysis
Analyze the Android project architecture and dependencies.

# Translation sync
Sync translations from Excel: ./translations/strings.xlsx

# Code review
Review my staged changes for Android crash risks and boundary conditions.

# APK analysis
Analyze this APK for permissions and security issues: ./build/outputs/apk/release/app-release.apk

Note: Cursor rules are static prompt injection and do not support hooks, slash commands, or multi-agent orchestration. The workflow knowledge is fully preserved, but automated triggers (e.g. Stop Hook in review-loop) are not available.

Using Skills in Gemini CLI

Pre-converted files for Gemini CLI are in the gemini-rules/ directory. Two integration methods are available.

Method 1 — GEMINI.md (Recommended)

Inject the skill content into your project's GEMINI.md using the @import syntax:

# In your Android project root
mkdir -p .gemini

# Append import references to your project GEMINI.md
cat >> GEMINI.md << 'EOF'

# Android Skills
@/path/to/android-claude-code-skills/gemini-rules/android-project-analyzer.md
@/path/to/android-claude-code-skills/gemini-rules/android-translation-sync.md
@/path/to/android-claude-code-skills/gemini-rules/android-change-review.md
@/path/to/android-claude-code-skills/gemini-rules/apk-analyzer.md
EOF

Or copy the files directly and reference them locally:

mkdir -p .gemini/skills
cp android-claude-code-skills/gemini-rules/*.md .gemini/skills/

# Then in GEMINI.md
cat >> GEMINI.md << 'EOF'
@.gemini/skills/android-project-analyzer.md
@.gemini/skills/android-translation-sync.md
@.gemini/skills/android-change-review.md
@.gemini/skills/apk-analyzer.md
EOF

Once added, simply describe your task naturally:

Sync translations from Excel: ./translations/strings.xlsx

Review my staged changes for Android crash risks.

Analyze this APK: ./build/outputs/apk/release/app-release.apk

Method 2 — Custom Slash Commands

Install the .toml command files to get /project-analyzer, /translation-sync, /change-review, and /apk-analyzer slash commands:

# Global install (available in all projects)
mkdir -p ~/.gemini/commands
cp android-claude-code-skills/gemini-rules/commands/*.toml ~/.gemini/commands/

# OR project-level install
mkdir -p .gemini/commands
cp android-claude-code-skills/gemini-rules/commands/*.toml .gemini/commands/

Then use directly in Gemini CLI:

# Analyze project
/project-analyzer

# Sync translations
/translation-sync ./translations/strings.xlsx

# Review staged changes
/change-review staged

# Review a specific commit
/change-review abc1234

# Analyze an APK
/apk-analyzer ./build/outputs/apk/release/app-release.apk

Capability Comparison

| Feature | Claude Code | Codex CLI | Gemini CLI | Cursor | |---|---|---|---|---| | Workflow knowledge | ✅ Full | ✅ Full | ✅ Full | ✅ Full | | Skill format | SKILL.md | SKILL.md (native) | GEMINI.md / .md | .mdc rules | | Skill trigger | /plugin + slash | $skill-name or auto | natural language | natural language | | Auto-detect from description | ✅ | ✅ (allow_implicit_invocation) | ✅ | ✅ | | Context file | CLAUDE.md | AGENTS.md | GEMINI.md | .cursor/rules/ | | Stop Hook / lifecycle hooks | ✅ | ❌ | ❌ | ❌ | | Multi-agent orchestration | ✅ | ❌ | ❌ | ❌ |

Adding Skills

Create your skills in the skills/ directory:

skills/
└── your-skill-name/
    ├── SKILL.md
    └── agents/
        └── openai.yaml   # Optional (recommended for Codex UI)

Each skill requires a SKILL.md file with:

YAML frontmatter - name and description (required for Codex skill triggering)
Description body - Workflow, rules, and reusable knowledge for the skill
Optional agent metadata - agents/openai.yaml for Codex UI display name/description/prompt

Minimal SKILL.md example:

---
name: your-skill-name
description: Describe what the skill does and when to use it so Codex can trigger it.
---

# Your Skill Name

Skill instructions...

Validation

Local validation (Codex-compatible SKILL.md + agents/openai.yaml):

python3 -m pip install --user PyYAML
python3 scripts/validate_skills.py

Pre-commit hook (run validation automatically before commit):

python3 -m pip install --user pre-commit
pre-commit install
pre-commit run --all-files

This repository also includes GitHub Actions CI at .github/workflows/validate-skills.yml to run the same checks on push/pull request.

Project Structure

android-claude-code-skills/
├── .claude-plugin/
│   ├── plugin.json         # Plugin manifest
│   └── marketplace.json    # Marketplace configuration
├── .github/
│   └── workflows/
│       ├── validate-skills.yml   # CI: validate all skills on push
│       └── publish-npm.yml       # CI: publish to npm on version tag
├── skills/                 # Available skills
│   ├── android-project-analyzer/
│   │   ├── SKILL.md
│   │   └── agents/openai.yaml
│   ├── android-translation-sync/
│   │   ├── SKILL.md
│   │   └── agents/openai.yaml
│   ├── android-change-review/
│   │   ├── SKILL.md
│   │   └── agents/openai.yaml
│   ├── apk-analyzer/
│   │   ├── SKILL.md
│   │   └── agents/openai.yaml
│   └── template/
│       ├── SKILL.md        # Codex-compatible skill template
│       └── agents/openai.yaml
├── cursor-rules/           # Cursor IDE rules (.mdc)
│   ├── android-project-analyzer.mdc
│   ├── android-translation-sync.mdc
│   ├── android-change-review.mdc
│   └── apk-analyzer.mdc
├── gemini-rules/           # Gemini CLI rules
│   ├── android-project-analyzer.md
│   ├── android-translation-sync.md
│   ├── android-change-review.md
│   ├── apk-analyzer.md
│   └── commands/           # Gemini slash commands (.toml)
│       ├── project-analyzer.toml
│       ├── translation-sync.toml
│       ├── change-review.toml
│       └── apk-analyzer.toml
├── llms.txt                # AI crawler discoverability
├── package.json            # npm registry metadata
├── AGENTS.md               # Agent instructions (OpenAI Codex)
├── CLAUDE.md               # Project overview (English)
├── CLAUDE_CN.md            # 项目概述（中文）
├── CLAUDE_JP.md            # プロジェクト概要（日本語）
├── README.md               # This file (English)
├── README_CN.md            # 说明文档（中文）
├── README_JP.md            # ドキュメント（日本語）
├── CHANGELOG.md            # Changelog (English)
└── CHANGELOG_JP.md         # 変更履歴（日本語）

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License

Published

Vulnerabilities

Links

Maintainers

Keywords

Readme

Android Claude Code Skills

Table of Contents

Repository Information

OpenAI Codex Compatibility

Installation

Updating Plugin

What Gets Updated

Verify Update Success

Available Skills

Android Project Analyzer

Android Translation Sync

Android Change Review

APK Analyzer

Third-Party Skills

review-loop — Automated Code Review Loop

claude-codex — Multi-AI Orchestration Pipeline

Using Skills in Codex CLI

Installation

Usage

Verify Installation

Using Skills in Cursor

Available Cursor Rules

Installation

Usage in Cursor

Using Skills in Gemini CLI

Method 1 — GEMINI.md (Recommended)

Method 2 — Custom Slash Commands

Capability Comparison

Adding Skills

Validation

Project Structure

Contributing

License