How does the hybrid routing mechanism work?

It uses fast, deterministic pattern matching for low-risk tiers and automatically escalates high-risk (Red/Critical) tasks to a 'war-room-checkpoint' for deeper reversibility analysis.

Does it integrate with existing task metadata?

Yes, it extends standard task objects with risk_tier and risk_reason fields, allowing downstream orchestrators to consume and act on risk data.

What are the four risk tiers used by this skill?

The skill uses Green (Safe/Trivial), Yellow (Caution/Module-level), Red (Danger/Cross-module), and Critical (Stop/Irreversible) to categorize task impact and required oversight.

Can this skill prevent accidental data loss?

Yes, by classifying destructive tasks as 'Critical', it enforces human approval and full reversibility scoring before an agent is permitted to proceed.

Risk Classification & Safety Gates

Name: Risk Classification & Safety Gates
Author: micsapp

bymicsapp

0•

安全与测试

Implements a four-tier safety model to classify AI agent tasks by risk level and enforce appropriate verification gates.

This skill provides a robust framework for assessing the operational risk of automated tasks before execution. By using a hybrid approach—combining fast heuristic file-pattern matching for low-risk tasks with deep reversibility scoring for critical operations—it ensures that AI agents operate within safe boundaries. It helps teams manage code safety by defining clear verification requirements, preventing irreversible data loss, and determining when human oversight or 'war-room' analysis is mandatory for complex system changes.

主要功能

010 GitHub stars

02Task metadata extensions to track risk profiles throughout the development lifecycle

03Integration with reversibility scoring (RS) for high-stakes code modifications

04Dynamic verification gates that adjust based on task impact and complexity

05Four-tier risk model (Green, Yellow, Red, Critical) for granular task assessment

06Heuristic-based file pattern matching for rapid, automated risk classification

使用场景

01Automating the routing of code changes based on whether they affect core infrastructure or peripheral documentation

02Preventing automated agents from executing irreversible database migrations without human approval

03Determining parallel execution safety between multiple concurrent AI-driven development tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add micsapp/micstec-skills risk-classification

For use in Claude.ai and ChatGPT

主要功能

010 GitHub stars

02Task metadata extensions to track risk profiles throughout the development lifecycle

03Integration with reversibility scoring (RS) for high-stakes code modifications

04Dynamic verification gates that adjust based on task impact and complexity

05Four-tier risk model (Green, Yellow, Red, Critical) for granular task assessment

06Heuristic-based file pattern matching for rapid, automated risk classification

使用场景

01Automating the routing of code changes based on whether they affect core infrastructure or peripheral documentation

02Preventing automated agents from executing irreversible database migrations without human approval

03Determining parallel execution safety between multiple concurrent AI-driven development tasks