agent-conduct

Enforce workspace boundaries, terminal safety, and Git hygiene for AI agents.

Updated Nov 27, 2025

Search Tags:#git #security #safety #workspace #terminal #agent conduct

Ready to run

npx skills add https://github.com/wtsi-hgi/llm-knowledge-base --skill agent-conduct

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-conduct
Source: https://github.com/wtsi-hgi/llm-knowledge-base/tree/main/.github/skills/agent-conduct
Command: npx skills add https://github.com/wtsi-hgi/llm-knowledge-base --skill agent-conduct

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill establishes essential safety and security guidelines for all AI agents, preventing accidental data loss, system compromise, and unwanted interactions with the development environment.

Core Features & Use Cases

Workspace Boundary Enforcement: Ensures all file operations remain strictly within the repository.
Terminal Safety Protocols: Prevents the execution of commands that could trigger security prompts or lead to unintended consequences.
Git Hygiene: Mandates safe Git practices to avoid accidental commits or history rewriting.
Use Case: Before any agent begins a task involving file modifications or terminal commands, this skill's rules are consulted to ensure all actions are safe and contained.

Quick Start

Follow the workspace boundary rules and never write outside the repository.

Frequently Asked Questions about agent-conduct

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I enforce AI agent security boundaries within a development workspace?▼

To prevent accidental data loss during git operations, this skill mandates safe Git practices and terminal command hygiene. It blocks commands that could trigger security prompts or rewrite history, ensuring safe repository modifications.

How does terminal command hygiene prevent prompt injection in AI agents?▼

Terminal command hygiene prevents prompt injection by blocking the execution of commands that could lead to unintended consequences or system compromise. It scrutinizes terminal operations to maintain a secure agent environment.

Do I need workspace boundary rules for AI agents modifying files in a repository?▼

Yes, workspace boundary rules are essential for AI agents modifying files. This skill enforces these boundaries to ensure all file modifications remain strictly within the repository, preventing accidental data loss and system compromise.

What are the limitations of relying on standard git safety for automated agent tasks?▼

Standard git safety lacks strict operational protocols for automated agent tasks. This skill addresses limitations by enforcing workspace boundaries, terminal command hygiene, and scratch work management to prevent data loss and prompt injection.