A robust framework for defining immutable behavioral rules, permission boundaries, and safety interlocks that AI agents cannot override.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install skylv-agent-constitution-guard
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 skylv-agent-constitution-guard 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 Agent Constitution Guard?
Agent Constitution Guard is a specialized safety system designed to provide AI agents with a set of enforceable, constitutional rules. By utilizing this within the Openclaw Skills framework, developers can establish multi-layered defense mechanisms that protect infrastructure from accidental or malicious actions. The skill ensures that agents operate within strict boundaries, preventing unauthorized data exfiltration, system modifications, or excessive API spending, even when faced with prompt injection or social engineering attempts.
This system acts as a foundational layer for responsible AI deployment, offering a structured way to manage agent permissions. Integrating these Openclaw Skills into your workflow allows for a higher level of trust and reliability, as every action is cross-referenced against a predefined constitution that remains immutable regardless of the context or user pressure.
Agent Constitution Guard Use Cases
- Prevent agents from executing destructive commands like deleting production databases.
- Limit API spending by enforcing daily budget caps and cost-per-call constraints.
- Safeguard sensitive PII data by blocking unauthorized exports in compliance with GDPR.
- Implement human-in-the-loop requirements for high-risk operations via escalation policies.
- Control file system access by restricting an agent to specific workspace directories.
How Agent Constitution Guard Works
- The developer initializes a constitution for the agent, creating a dedicated directory for rules and logs.
- Specific rules are defined across different levels, from immutable safety boundaries to advisory performance hints.
- Before executing any action, the agent calls the guard check to verify the request against the active constitution.
- If a rule violation is detected at the immutable level, the action is automatically blocked; if it triggers an owner-only rule, an escalation request is sent.
- The system records every decision in an immutable audit trail, providing a transparent history of agent behavior for security reviews.
Agent Constitution Guard Setup
To get started with this skill for Openclaw Skills, follow these installation steps:
- Initialize the constitution directory:
node constitution.js init --name "my-agent" --owner "[email protected]"
- Add an immutable safety rule:
node constitution.js rule add --level immutable --action deny --scope "external_write" --description "Block all unauthorized external API writes"
- Integrate the check into your application logic:
const decision = guard.check('external_write', { target: 'api.example.com' });
if (!decision.allowed) { /* Handle restriction */ }
Agent Constitution Guard Data Schema & Taxonomy
The skill manages its data within a .constitution/ directory, using a structured format to ensure high-speed lookups and audit integrity:
| File | Role | Format |
|---|---|---|
rules.json |
Stores the active hierarchy of behavioral constraints | JSON |
audit.log |
An append-only log of every permission check performed | JSONL |
config.json |
Global settings including escalation timeouts and retention policies | JSON |
Each rule entry includes an ID, scope, level (Immutable, Owner-only, Mutable, or Advisory), and specific conditions such as environment or data type restrictions.
Agent Constitution Guard Advanced Features
- Multi-layer permission architecture that separates hard safety boundaries from flexible performance hints.
- Support for human-in-the-loop escalation via webhooks, Slack, or email for sensitive operations.
- Immutable audit logs that can be exported in CSV format for compliance reporting and security auditing.
- CI/CD integration tools to verify constitutional safety during the deployment pipeline.
- Context-aware rule conditioning that adjusts permissions based on real-time environmental variables.
SKILL.md
Loading
Related Openclaw Skills
Liuyao Divination Agent
An AI-powered traditional Chinese divination skill that combines ancient I Ching wisdom with Python-driven hexagram generation.
Agent Security Harness
A comprehensive security testing harness providing 470 executable protocol and behavioral tests for AI agent systems.
Kleinanzeigen Assistant Skill
An intelligent automation skill for generating, researching, and staging classified ad drafts on kleinanzeigen.de.
Liuyao Divination
A sophisticated AI divination tool providing I Ching readings through virtual coin simulation and cross-cultural philosophical analysis.
Agent Settings Manager
A comprehensive configuration management system for AI agents to handle environment variables, secrets, and versioned settings.
Agent Evaluator
A professional evaluation framework that scores AI agent performance across five critical dimensions to provide actionable improvement insights.
Featured*
