A professional-grade configuration skill for managing tool access policies, execution security, and agent capabilities within the Openclaw ecosystem.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install configure-tools
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 configure-tools 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 Configure Tools?
The Configure Tools skill provides a robust framework for managing how AI agents interact with their environment. It enables developers to define granular access control through profile-based permissions, explicit tool allow/deny lists, and sophisticated execution security layers. By utilizing this skill, you ensure that every agent operates within a defined safety boundary, preventing unauthorized code execution or data access.
Built to integrate seamlessly with Openclaw Skills, this configuration utility handles everything from global policy layering to provider-specific restrictions. It is the essential tool for maintaining the principle of least privilege across complex multi-agent deployments, allowing you to lock down sensitive capabilities while enabling specific plugin tools for specialized tasks.
Configure Tools Use Cases
- Setting up tool access for a new AI agent with specific role requirements.
- Restricting shell command execution to prevent security vulnerabilities.
- Configuring execution security levels such as sandboxing or gateway-based approvals.
- Enabling specific plugin tools like web search or task management via additive policies.
- Applying different tool profiles based on the specific LLM provider or model being used.
How Configure Tools Works
- Scope Identification: Determine if the configuration applies globally, to a specific agent ID, or to a specific model provider.
- Profile Selection: Assign a standard tool profile (coding, messaging, minimal) to establish a baseline of capabilities.
- Policy Refinement: Use group shorthands to fine-tune access by explicitly allowing or denying specific toolsets.
- Security Hardening: Define the execution environment (host) and approval requirements (ask) for shell operations.
- Deployment: Apply the generated configuration patch via the gateway tool or direct modification of the Openclaw configuration file.
Configure Tools Setup
To configure tools using this skill, you can apply patches directly through the Openclaw gateway.
# Example of applying a configuration patch via CLI
openclaw gateway config.patch --patch '{"tools": {"profile": "coding"}}'
Alternatively, manually edit your local configuration file:
# Open the configuration file
nano ~/.openclaw/openclaw.json
# Restart the Gateway to apply changes
openclaw restart
Configure Tools Data Schema & Taxonomy
The skill manages tool policies using a structured JSON schema within the Openclaw environment:
| Field | Description | Type |
|---|---|---|
profile |
Predefined tool sets (full, coding, messaging, minimal) | String |
allow / deny |
Explicit lists of tool names or group shorthands | Array |
alsoAllow |
Additive permissions for plugin tools | Array |
exec |
Configuration for host, security, and approval flow | Object |
byProvider |
Model-specific tool overrides | Map |
Configure Tools Advanced Features
- Policy Layering: Inherit global tool rules while overriding specific settings at the agent or model level.
- Execution Sandboxing: Isolate shell command execution using the
sandboxhost for maximum security. - Group Shorthands: Manage large sets of tools (e.g.,
group:ui,group:runtime) with single-line declarations. - Conditional Approvals: Configure
ask: "on-miss"to require human intervention only when an agent attempts to use an unauthorized tool. - Provider-Specific Optimization: Automatically downgrade tool access for less capable models to ensure stability.
SKILL.md
Loading
Related Openclaw Skills
Site Inspector
An offline-first reference tool providing instant access to DevTools best practices, implementation patterns, and debugging guides.
17CE Global Website Speed Test
A real-time global performance monitoring tool that tracks website speed, HTTP response times, and AI model availability via 17CE's distributed node network.
SilicaClaw Network Config
A specialized skill for inspecting and modifying SilicaClaw runtime network modes and public discovery states.
WeChat Video Downloader
A specialized automation tool for extracting and saving embedded video files from WeChat Official Account articles.
Qwen Image Skill
A professional AI image generation and editing skill utilizing the Qwen model via DashScope API for seamless text-to-image and image-to-image workflows.
System Hardener
A comprehensive local reference tool providing instant access to system hardening concepts, implementation patterns, and security best practices.
Featured*
