What is the primary purpose of the sdd-guardrails skill?

It generates a guardrails.md file that defines the safety limits, permission boundaries, and approval requirements for an AI agent to prevent unintended or harmful actions.

Can I define custom resource limits for the AI agent?

Yes, the generated guardrails include sections for maximum execution time, file size limits, and API call quotas which can be customized per project.

What prerequisites are needed to run this skill?

The skill requires that requirements.md and threat-model.md files already exist in the target directory, as it uses these to inform the specific guardrail definitions.

How does this help with auditing?

The skill defines a mandatory Audit Trail section, ensuring that all tool calls, file changes, and human approvals are logged in a structured format for later review.

What are Human-in-the-Loop gates?

These are specific operational checkpoints where the AI agent is required to stop and wait for explicit human approval before proceeding with a task, such as a code deployment or a large-scale deletion.

AI Agent Guardrails Generator

Name: AI Agent Guardrails Generator
Author: taiyousan15

bytaiyousan15

0•

安全与测试

Generates comprehensive security guardrails and safety policies for AI agents to ensure controlled and auditable operations.

The sdd-guardrails skill is designed to secure AI agent workflows by generating a standardized guardrails.md file based on existing requirements and threat models. It establishes clear permission boundaries, resource limits, and mandatory Human-in-the-Loop (HITL) approval gates for sensitive operations such as production deployments or database modifications. By automating the creation of these safety protocols, developers can ensure that Claude and other AI agents operate within strict ethical and technical constraints while maintaining a full audit trail of all activities.

主要功能

010 GitHub stars

02Automated generation of guardrails.md from requirements and threat models

03Standardizes audit trail requirements and error handling policies

04Establishes mandatory Human-in-the-Loop approval gates for high-risk tasks

05Defines strict file system and command execution permission boundaries

06Configures resource limits for execution time, file sizes, and API usage

使用场景

01Enforcing safety protocols for autonomous AI agents in production environments

02Implementing governance and compliance standards for AI-driven development

03Defining granular permission sets for AI tools interacting with sensitive codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add taiyousan15/taisun_agent sdd-guardrails

For use in Claude.ai and ChatGPT

Download Skill