safeclaw 0.1.0

SafeClaw - Keeping Your Claw Safe. Real-time monitoring and security for OpenClaw AI agents.

SafeClaw

Keeping Your Claw Safe.

Real-time monitoring, security scanning, and red team testing for OpenClaw AI agents.

---

What is SafeClaw?

SafeClaw is a security-focused companion for OpenClaw AI agents. It provides a unified dashboard to monitor agent activity, scan system assets, and perform automated red team testing — all from a single safeclaw start command.

Core Modules

Module	Description
Claw Monitor	Real-time session timeline with event tracking, token usage, and tool call inspection
Safe Chat	Secure gateway to chat with your OpenClaw agent through a managed interface
Asset Shield	Hardware inventory, file system scanning, software audit, and security risk assessment
Red Teaming	Automated multi-turn attack simulation — select a category, generate decomposed attacks, and execute them against a live agent
Onboard Setup	Interactive wizard to install and configure OpenClaw CLI with full PTY support

---

Installation

Option A: Install from GitHub (recommended)

pip install git+https://github.com/dyf-2316/SafeClaw.git

Option B: Clone and install locally

git clone https://github.com/dyf-2316/SafeClaw.git
cd SafeClaw
pip install .

Option C: Development install

git clone https://github.com/dyf-2316/SafeClaw.git
cd SafeClaw
pip install -e ".[dev]"

Requires Python 3.11+. The frontend is pre-built and bundled in the package — no Node.js needed for production use.

---

Quick Start

safeclaw start

Browser opens automatically at http://127.0.0.1:6874. Database is created at ~/.safeclaw/data.db on first launch.

CLI Reference

Usage: safeclaw [OPTIONS] COMMAND [ARGS]...

Commands:
  start    Start the SafeClaw server
  version  Show SafeClaw version

Options for `safeclaw start`:
  -p, --port INTEGER       Server port              [default: 6874]
  -h, --host TEXT          Bind address             [default: 127.0.0.1]
      --no-browser         Don't open browser automatically
      --reload             Enable auto-reload (dev mode)

Examples:

safeclaw start                          # default settings
safeclaw start --port 8080              # custom port
safeclaw start --host 0.0.0.0           # accessible from LAN
safeclaw start --no-browser --reload    # headless dev mode

---

Development Setup

For contributing or modifying SafeClaw, run the backend and frontend as separate processes with hot reload.

Prerequisites

• Python 3.11+
• Node.js 18+
• uv (recommended) or pip

1. Clone & Install Backend

git clone https://github.com/dyf-2316/SafeClaw.git
cd SafeClaw

uv venv
uv pip install -e ".[dev]"

2. Install Frontend Dependencies

cd frontend
npm install
cd ..

3. Configure Environment (optional)

cp .env.example .env

Defaults work out of the box. Edit .env only if you need to change ports or paths.

4. Start Backend (Terminal 1)

source .venv/bin/activate
python run.py

Backend runs at http://localhost:6874 with auto-reload enabled.

5. Start Frontend (Terminal 2)

cd frontend
npm run dev

Frontend runs at http://localhost:3000 with HMR. API calls are proxied to the backend automatically.

6. Build Frontend into Package

cd frontend
npm run build

Outputs to src/safeclaw/static/. After building, safeclaw start serves the embedded frontend directly.

---

Architecture

                  Browser
                    |
            :6874 (production)
            :3000 (dev, proxied)
                    |
        +-----------+-----------+
        |     FastAPI Server    |
        |                       |
        |  /api/*   REST APIs   |
        |  /*       Static SPA  |
        +-----------+-----------+
                    |
        +-----------+-----------+
        |                       |
   SQLite DB           OpenClaw Sessions
 ~/.safeclaw/           ~/.openclaw/
   data.db            agents/main/sessions/

Tech Stack

Layer	Technology
Backend	Python 3.11, FastAPI, SQLAlchemy (async), uvicorn
Frontend	React 19, TypeScript, Vite, Tailwind CSS 4
Database	SQLite (via aiosqlite)
CLI	Typer + Rich
File Sync	Watchdog (real-time JSONL parsing)

---

Project Structure

SafeClaw/
├── src/safeclaw/                  # Python package
│   ├── cli.py                     # CLI entry point (safeclaw start)
│   ├── config.py                  # Settings (pydantic-settings)
│   ├── database.py                # SQLite async engine
│   ├── gateway_client.py          # OpenClaw gateway client
│   ├── api/
│   │   ├── main.py                # FastAPI app + static serving
│   │   └── routes/
│   │       ├── sessions.py        # Session CRUD
│   │       ├── events.py          # Event timeline
│   │       ├── messages.py        # Message history
│   │       ├── stats.py           # Token & usage stats
│   │       ├── assets.py          # Hardware & file scanning
│   │       ├── redteam.py         # Red team attack generation
│   │       ├── chat.py            # Agent chat gateway
│   │       ├── system.py          # OpenClaw install/onboard (PTY)
│   │       ├── guard.py           # AgentDoG safety guard
│   │       └── trace.py           # Trace inspection
│   ├── models/                    # ORM models (Session, Message, Event, ToolCall)
│   ├── services/                  # Background sync & stats
│   ├── asset_scanner/             # System asset scanner
│   └── static/                    # Built frontend (auto-generated)
├── frontend/                      # React SPA
│   ├── src/
│   │   ├── pages/                 # Monitor, Chat, Assets, RiskScanner, Setup, Home
│   │   ├── components/            # Layout, shared UI
│   │   └── services/api.ts        # Axios API client
│   └── vite.config.ts
├── external/                      # External tools (RedWork data)
├── pyproject.toml                 # Package metadata
├── run.py                         # Dev server script
└── .env.example                   # Configuration template

---

Configuration

SafeClaw reads settings from environment variables or a .env file:

Variable	Default	Description
DATABASE_URL	~/.safeclaw/data.db	Database path (auto-created)
OPENCLAW_SESSIONS_DIR	~/.openclaw/agents/main/sessions	OpenClaw session JSONL directory
API_HOST	0.0.0.0	Server bind address
API_PORT	6874	Server port
LOG_LEVEL	INFO	Logging level (DEBUG, INFO, WARNING, ERROR)
ENABLE_FILE_WATCHER	true	Auto-watch and sync session files
WATCH_INTERVAL_SECONDS	1	File watcher polling interval
DATA_DIR	~/.safeclaw	Data directory for DB and config

---

API Overview

All endpoints are prefixed with /api. Full OpenAPI docs available at http://localhost:6874/docs when running.

Prefix	Description
/api/sessions	List, inspect, and delete agent sessions
/api/events	Query interaction events with timing and stats
/api/messages	Browse messages with content and token info
/api/stats	Aggregated stats by model, daily usage, overview
/api/assets	Hardware scan, file scan, software audit, safety check
/api/redteam	List instructions, generate decomposed attacks
/api/chat	Start sessions, send messages to OpenClaw agent
/api/system	OpenClaw status, install, onboard (PTY streaming)
/api/guard	AgentDoG safety check for sessions
/api/trace	Trace and inspect agent execution

---

License

MIT