Skill#buildwithclaude #ai

Sosa governance

Gerçek zamanlı SOSA yönetişim uygulaması — Denetim politikalarını, güven derecelerini ve yetenek sınırlarını, yürütülürken AI aracı eylemlerine uygulayın.

yazan kuscubasiece·8 Eylül 2026

Kaynak: davepoon/claude-code-subagents-collection · MIT

Kurulum

mkdir -p .claude/skills/sosa-governance && curl -sL https://www.antrophiclove.com/api/kaynak/sosa-governance/raw -o .claude/skills/sosa-governance/SKILL.md

Hedef dosya: .claude/skills/sosa-governance/SKILL.md

Önizleme Ham metin ⑂ Kendi sürümünü oluştur

---
name: sosa-governance
description: "Real-time SOSA governance enforcement — apply supervision policies, trust gradients, and capability boundaries to AI agent actions as they execute."
category: quality-security
---

# SOSA Governance Engine

Runtime enforcement layer for the SOSA framework. Intercepts agent actions and applies governance rules before execution.

## Capabilities

- Trust gradient: first-time actions get full confirmation, repeat actions are expedited
- Capability boundaries: restrict what each skill/agent can access
- Policy enforcement: block dangerous patterns (hardcoded secrets, unbounded actions)
- Audit trail: log all governance decisions for review

## Setup

No API keys required. Install from `MSApps-Mobile/claude-plugins` marketplace.