AndroClaw CLAUDE.md

Repository instructions for AndroClaw, an Android assistant that connects language models to actions on a phone.

In plain words
What is it for?
Guiding development of an Android app that can send messages, make calls, search the web, read files, take notes, and control other apps.
Why use it?
They give agents the project’s architecture, technologies, folder layout, and supported language-model providers before code changes are made.

Instructions file

Install

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.

agentmods
npx agentmods add instructions/androclawd/androclaw/claude-md
Clone the repo
git clone --depth 1 https://github.com/AndroClawD/AndroClaw
Per session 1,831 This file is loaded in full into every session.
When invoked 1,831 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.01831 $0.01831
Opus 5 $0.00915 $0.00915
Sonnet 5 $0.00366 $0.00366
Haiku 4.5 $0.00183 $0.00183

Measured 3d ago against content hash df3d0faaf164, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

AndroClaw CLAUDE.md 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 3d 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.

CLAUDE.md · 159 lines

How it starts

The opening of the file, as written. The whole thing — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AndroClaw

Built by Pranav Patil

AndroClaw is a powerful AI-powered Android assistant that runs directly on your device. It connects to multiple LLM providers and executes real actions on your phone — from sending messages and making calls to searching the web, reading files, taking notes, and controlling other apps.

Architecture

  • Language: Kotlin
  • UI: Jetpack Compose + Material 3
  • DI: Hilt (Dagger)
  • Database: Room (SQLite) with versioned migrations
  • Networking: OkHttp + Retrofit + Jsoup
  • API: Multi-provider LLM support with streaming
  • Min SDK: 29 (Android 10)
  • Target SDK: 35

Project Structure

app/src/main/java/com/androclaw/
├── api/
│   ├── provider/           # Multi-LLM provider abstraction
│   │   ├── LlmProvider.kt       # Provider interface + streaming types
│   │   ├── ClaudeProvider.kt     # Anthropic Claude (SSE streaming)
│   │   ├── OpenAIProvider.kt     # OpenAI + compatible APIs (Groq, OpenRouter)
│   │   ├── GeminiProvider.kt     # Google Gemini
│   │   └── ProviderRegistry.kt   # Provider management
│   ├── ClaudeRepository.kt      # Agentic loop, tool orchestration, streaming
│   ├── ClaudeApiService.kt      # Retrofit interface (legacy, kept for compat)
│   ├── ClaudeResponseParser.kt  # Custom JSON parser
│   ├── ToolDefinitions.kt       # All tool schemas (30+ tools)
│   └── models/ClaudeModels.kt   # Request/response DTOs
├── db/
│   ├── AndroClawDatabase.kt     # Room DB with migrations v2→v5
│   ├── ConversationEntity.kt    # Chat conversations
│   ├── MessageEntity.kt         # Chat messages
│   ├── MemoryEntity.kt          # Persistent AI memory
│   ├── NoteEntity.kt            # User notes
│   └── SkillEntity.kt           # Custom slash command skills
├── di/
│   └── AppModule.kt             # Hilt dependency injection
├── service/
│   ├── AndroClawAccessibilityService.kt  # UI automation + screenshots
│   ├── FloatingButtonService.kt          # Overlay floating button
│   ├── NotificationReaderService.kt      # Read notification content
│   ├── OverlayChatManager.kt             # Overlay chat state
│   └── BootReceiver.kt                   # Auto-start on boot
├── tools/                    # 30+ tool handlers
│   ├── ToolExecutor.kt              # Tool router/dispatcher
│   ├── WebSearchToolHandler.kt      # Web search (DuckDuckGo + Google)
│   ├── WebFetchToolHandler.kt       # URL content reader
│   ├── MemoryToolHandler.kt         # Persistent memory across sessions
│   ├── NotesToolHandler.kt          # Local note-taking
│   ├── SkillToolHandler.kt          # Custom slash commands
│   ├── FileToolHandler.kt           # File explorer (find/read/tree/grep/glob)
│   ├── LocationToolHandler.kt       # GPS + reverse geocoding
│   ├── SmsReaderToolHandler.kt      # Read SMS history
│   ├── CallLogToolHandler.kt        # Call history
│   ├── ScreenshotToolHandler.kt     # Screenshot + vision analysis
│   ├── TtsToolHandler.kt            # Text-to-speech
│   ├── AppLaunchToolHandler.kt      # App launcher with fuzzy matching
│   ├── AutoScrollToolHandler.kt     # Reels/Shorts auto-scroll
│   └── ... (15+ more handlers)
├── ui/
│   ├── ChatScreen.kt + ChatViewModel.kt
│   ├── SettingsScreen.kt + SettingsViewModel.kt
│   ├── OnboardingScreen.kt
│   └── components/ (InputBar, MessageBubble, ChatHistoryDrawer, etc.)
└── utils/
    ├── Constants.kt              # System prompt, prefs keys, tool names
    ├── PermissionHelper.kt       # Runtime permission management
    └── VoiceInputManager.kt      # Speech-to-text

Read the full file on GitHub · 159 lines

Changes

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.

  1. 3d ago First seen · 159 lines · 1,831 tokens per session scan A df3d0faaf164

Subscribe to this mod's changes

AndroClaw CLAUDE.md is an instructions file published in the GitHub repository AndroClawD/AndroClaw (47 stars, last pushed 4mo ago), licensed MIT. It adds 1,831 tokens to every session, about $0.0092 per session on Opus 5. 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.