What is the principle of least privilege in this skill?

It involves granting agents only the minimum permissions (claims) necessary to perform their specific task, which significantly reduces security risks during automated operations.

Does this support multi-agent workflows?

Absolutely. It is designed specifically to manage permissions and coordination between different agents to ensure secure handoffs and operations.

Can I restrict an agent to a specific folder?

Yes, using Scope Patterns like '/src/**' or '/config/*.toml', you can limit an agent's claims to specific directories or file types.

How do I verify if an agent has a specific permission?

You can use the 'claims check' command to instantly verify if a specific agent holds a particular claim, such as 'write' or 'execute', before proceeding.

What are the available security levels?

The skill supports predefined levels: 'minimal' (read-only), 'standard' (read/write/execute), 'elevated' (adds spawn/memory), and 'admin' (all claims).

Claims-Based Authorization

Name: Claims-Based Authorization
Author: nbossn

bynbossn

0•

보안 및 테스팅

Manages granular, claims-based permissions for secure multi-agent coordination and resource access control.

The Claims-Based Authorization skill provides a robust framework for managing permissions across multi-agent workflows within Claude Code. It enables developers to implement the principle of least privilege by granting, revoking, and verifying specific access rights such as file reading, writing, command execution, and network access. With built-in support for scoped patterns and predefined security levels, it ensures that autonomous agents operate within safe, defined boundaries, making it essential for complex automation tasks where strict security and access control are paramount.

주요 기능

01Scoped resource patterns to limit access to specific directories or files

02Comprehensive auditing via claim listing and revocation commands

030 GitHub stars

04Multi-agent coordination with secure permission verification

05Granular claim management for read, write, execute, and admin actions

06Predefined security levels ranging from minimal to administrative access

사용 사례

01Restricting autonomous agents to specific project directories for security

02Coordinating multiple AI agents with tiered permission levels for complex tasks

03Implementing safe execution environments for AI-driven command-line operations

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nbossn/claude-code-base claims

For use in Claude.ai and ChatGPT