secure-openclaw
Run a 24/7 AI assistant on WhatsApp, Telegram, or Signal with persistent memory and tool access.
Connect Claude or Opencode to WhatsApp, Signal, and Telegram with persistent memory and app workflows.
Secure OpenClaw turns messaging apps like WhatsApp, Telegram, and Signal into an always-on AI assistant interface. It lets you run Claude or open-source models with tool execution, cron reminders, and 500+ app connections directly from your phone. Mental health practitioners can use it to manage practice schedules, administrative notes, and personal tasks via chat.
What it is
The tool connects AI models directly to your messaging apps through a self-hosted gateway daemon. It provides persistent memory across conversations, scheduled triggers for reminders, and file or command-line tool access. Channel security is enforced through explicit user and group allowlists to drop unauthorized messages.
Who it's for
It suits tech-savvy solo practitioners and practice managers who want an automated assistant on their daily messaging tools. It is not suitable for handling sensitive patient health information or protected clinical notes due to third-party API exposure and standard messaging protocol risks. Non-technical users without server administration experience should avoid it.
Key features
Multi-platform messaging gateway
Runs on WhatsApp, Telegram, Signal, and iMessage with sender allowlists to filter unauthorized inbound chats.
Dual model provider engine
Supports Anthropic's Claude models via Claude Code alongside open-source engines like Opencode.
Persistent memory system
Stores context and long-term facts across chat sessions to maintain continuity across messaging platforms.
Scheduled cron triggers
Sets reminders and automated background tasks directly from conversational commands.
Composio app ecosystem
Integrates with over 500 external apps including calendar tools, file stores, and communication services.
Up and running in 3 steps
Our honest verdict
What's great
- Direct control over channel allowlists prevents unauthorized users from querying your assistant.
- Combines local command execution with external app integrations using Composio.
- Supports persistent background memory so you do not lose context between messaging sessions.
- Can be switched between Anthropic models and open-source Opencode backends.
Where it falls short
- Requires hosting your own server or Docker container with ongoing API usage fees.
- Not HIPAA-compliant out of the box, making it unsafe for direct patient health data processing.
- Setup demands terminal command proficiency, environment variable management, and server administration.
- WhatsApp authentication requires periodic QR code re-linking and session storage maintenance.
Pricing & license
The codebase is free to use under the MIT license. You must pay for your own server hosting (typically $6 per month for a basic Linux VPS) and any API usage from Anthropic or Composio.
Open remote AI-training roles
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Can I use Secure OpenClaw for direct patient communication?
No, standard messaging protocols and third-party API integrations do not meet clinical compliance or HIPAA regulations for handling protected health information.
What hardware or server is required to run this?
A basic Linux VPS with at least 1 GB RAM and 2 GB swap space running Docker is sufficient to run the gateway.
Does it support voice messages or file attachments?
The assistant reads text inputs, executes terminal tools, and interfaces with connected app integrations through written prompts.
