An automated utility to install, configure, and verify the SafeClaw safety proxy for secure AI agent model routing.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install safeclaw-proxy
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install safeclaw-proxy using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is safeclaw-proxy?
This skill provides an automated end-to-end setup for the SafeClaw safety proxy, a critical component for developers using Openclaw Skills to monitor and secure LLM traffic. It handles environment detection, containerized deployment via Docker or Podman, and automatic routing reconfiguration for various agent modes.
By funneling model calls through a local or hosted proxy, users gain full visibility into their AI interactions through a dedicated dashboard while enforcing safety guardrails. The skill intelligently manages configuration patches for gateways, agent model files, and shell environments, ensuring that security is seamlessly integrated into your professional Openclaw Skills development workflow.
safeclaw-proxy Use Cases
- Setting up a local safety proxy for LLM traffic monitoring and auditing.
- Routing Openclaw Skills model traffic through a remote hosted SafeClaw instance.
- Debugging why specific model calls are not appearing in your monitoring dashboard.
- Automating the installation of prerequisites like Python 3.12 and the uv package manager.
- Enforcing real-time safety blocks on potentially harmful AI model requests.
How safeclaw-proxy Works
- Inspects the current environment to determine if Openclaw Skills is routing traffic via gateway config, agent model files, or shell variables.
- Performs pre-flight checks to verify localhost network permissions and elevated execution capabilities.
- Deploys the SafeClaw proxy container using Docker or Podman, with an automated fallback to a native Python installation if needed.
- Configures the routing path by patching provider settings or editing the agent models.json file to point at the new proxy URL.
- Executes a health check and a deliberate safety-triggering test call to verify that traffic is being correctly intercepted and logged.
safeclaw-proxy Setup
To initiate the setup within your Openclaw Skills environment, ensure you have an active session and necessary permissions for container management.
Containerized Deployment
The skill prefers running via Docker or Podman:
docker run -d --name safeclaw-proxy -p 8899:8899 ghcr.io/aceteam-ai/aep-proxy:latest
Manual Installation Fallback
If container runtimes are unavailable, the skill will attempt a local installation:
uv pip install aceteam-aep[all]
aceteam-aep proxy --port 8899 > /dev/null 2>&1 &
safeclaw-proxy Data Schema & Taxonomy
The skill manages and interacts with several data structures to ensure proper routing within Openclaw Skills:
| Component | Path / Key | Description |
|---|---|---|
| Gateway Config | models.providers.<provider>.baseUrl |
Re-points the provider to the local proxy v1 endpoint. |
| Agent Metadata | ~/.openclaw/agents/main/agent/models.json |
Updates specific agent routing configurations. |
| Shell Profile | OPENAI_BASE_URL |
Configures the environment variable for CLI-launched processes. |
| State API | {PROXY_URL}/aep/api/state |
Retrieves the current proxy state and call count data. |
safeclaw-proxy Advanced Features
- Intelligent host port fallback logic that cycles from 8899 down to 8895 to avoid port conflicts.
- Automated dependency management that handles Python 3.12 and uv installation on macOS and Linux systems.
- Configuration safety features including automatic backups of models.json before any modifications occur.
- Support for both local and hosted proxy modes to accommodate various Openclaw Skills deployment architectures.
- Integrated safety testing that interprets HTTP status codes to confirm active protection and dashboard logging.
SKILL.md
Loading
Related Openclaw Skills
AI Video Editor Kannada
A cloud-powered AI video editor specifically optimized for generating Kannada subtitles and professional video edits from raw footage.
Customer Memory Skill
A specialized skill that gives AI agents persistent memory of customer interactions and preferences using BlueColumn integration.
AI Video Generator Free Envato
A professional-grade AI tool that converts product images and logos into high-quality 1080p videos with automated music and text overlays.
TruClaw Biometric AI Guardrail
TruClaw is a biometric security plugin that intercepts sensitive AI tool calls and requires Face ID verification via a paired iOS device before execution.
Image to Video HD Free
An AI-powered skill that transforms static images into high-definition videos with smooth transitions and professional effects in under 60 seconds.
Instagram Video Editor App Download
A cloud-powered AI video editing skill that transforms raw clips into 1080p Instagram-ready content through simple text commands.
Featured*
