Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
A Python decorator layer that adds content filtering to agent outputs. Implements a Nano-Gate that checks decisions before execution and an Understanding Heart module that attaches remediation suggestions when risky content is detected. Designed to wrap existing agent functions without requiring structural changes to the underlying code.
The decorator pattern lets you add filtering to any existing agent function with a single annotation instead of restructuring the codebase.
When to use it
- Adding content filtering to an existing LLM agent without rewriting it
- Flagging potentially harmful outputs in an automated response pipeline
- Attaching remediation notes to risky agent actions before delivery
- Running agent decisions through a deterministic ethical gate at runtime
- Auditing agent outputs for balance checks before they reach end users
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: wrap an existing agent reply function with ethical output filtering
- 1Import validate_decision and guardian_wrap from integration_api
- 2Apply @guardian_wrap decorator to the existing generate_reply function
- 3Agent invokes the function, Nano-Gate checks the output against its filter rules
- 4Understanding Heart attaches suggestions if risky content is detected
- 5Wrapped response is returned with gate verdict and any attached annotations
Agent reply with Nano-Gate pass/flag verdict and optional remediation suggestions attached