posture

Define AI agent action scopes as scout, defense, or offense modes.

Updated Mar 13, 2026

Search Tags:#security #scope #control #action #mode #posture

Ready to run

npx skills add https://github.com/SunrisesIllNeverSee/moses-governance-cowork --skill posture

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: posture
Source: https://github.com/SunrisesIllNeverSee/moses-governance-cowork/tree/main/skills/posture
Command: npx skills add https://github.com/SunrisesIllNeverSee/moses-governance-cowork --skill posture

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill allows you to precisely control the scope of Claude's actions within a conversation, preventing unintended consequences and ensuring operations align with your current objectives, whether it's cautious analysis or decisive action.

Core Features & Use Cases

Define Operational Scope: Set Claude to 'scout' (read-only), 'defense' (protect existing state), or 'offense' (full action scope).
Contextual Enforcement: Ensures Claude adheres to the chosen posture for all subsequent actions.
Use Case: Before initiating a sensitive system update, you can set the posture to 'defense' to ensure Claude only performs actions that protect the current state and avoids any modifications.

Quick Start

Set the operational posture to scout mode.

Frequently Asked Questions about posture

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

FAQPage Schema

How do I control the operational scope of an AI agent during a sensitive task?▼

To control the operational scope of an AI agent, you can define distinct execution modes such as read-only reconnaissance, protective state preservation, or full-scope execution. This ensures operations align with your current objectives.

Can I restrict an AI to read-only reconnaissance before executing system modifications?▼

Yes, you can restrict an AI to read-only reconnaissance by setting its operational posture to scout mode. This prevents modifications during cautious analysis before you switch to full action scope.

What is the best way to prevent unintended consequences when an AI modifies a sensitive system?▼

The best way to prevent unintended consequences is to set the AI's posture to defense mode. This enforces protective state preservation, ensuring it only performs actions that protect the current state and avoids any modifications.

How does setting an action mode affect subsequent AI behavior in a conversational session?▼

Setting an action mode provides contextual enforcement for subsequent AI behavior within a conversational session. It ensures Claude strictly adheres to the chosen operational constraints, whether cautious analysis or decisive action.

When do I need to enforce action constraints on an AI in dynamic environments?▼

You need to enforce action constraints on an AI in dynamic or sensitive environments whenever cautious analysis or decisive action is required. Defining the operational posture prevents unintended consequences and manages the AI's action scope.

Does this posture Skill require any dependencies to manage AI behavior?▼

No, the posture Skill requires no dependencies to manage AI behavior. It independently defines the operational scope and action constraints for an AI agent within a conversational session.