skills.rest
Forexgod21
Forexgod21COMMUNITY
Explore all skills by this author →
WebsiteGitHub

utility-drift-detector

Detect and measure post-deployment drift in AI utility functions.

6|1|Updated Mar 5, 2026
Search Tags:#analysis#ai#drift-detection#drift#baseline#post-deployment#value-alignment
One-click install
npx skills add https://github.com/Forexgod21/YVYC-Claude-Skills --skill utility-drift-detector
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: utility-drift-detector
Source: https://github.com/Forexgod21/YVYC-Claude-Skills/tree/main/agentic/utility-drift-detector
Command: npx skills add https://github.com/Forexgod21/YVYC-Claude-Skills --skill utility-drift-detector

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Drift in AI system values and optimization targets after deployment can go unnoticed because outputs may still appear correct even when underlying objectives have shifted. This skill surfaces, quantifies, and helps remediate post-deployment drift by reconstructing the baseline, analyzing current outputs, classifying drift signals, and guiding responses.

Core Features & Use Cases

  • Baseline reconstruction to capture deployment intent and acceptable behavior.
  • Current output pattern analysis to detect shifts in language, tone, or decision patterns.
  • Drift signal classification across value weighting, proxy substitution, scope creep, constraint erosion, and feedback loop drift.
  • Magnitude assessment and root-cause analysis to enable actionable remediation and ongoing monitoring.

Quick Start

Install the skill and provide deployment baseline and current outputs to generate a drift report.

Frequently Asked Questions about utility-drift-detector

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

FAQPage Schema
How do I detect AI value drift after deployment?▼

AI value drift occurs when an AI system's underlying optimization targets shift post-deployment while outputs still appear correct. It surfaces through proxy substitution, scope creep, value weighting changes, constraint erosion, and feedback loop drift in live systems.

How do I measure drift magnitude in a live AI model?▼

Drift signal classification categorizes post-deployment AI behavioral shifts into value weighting, proxy substitution, scope creep, constraint erosion, and feedback loop drift. It determines the specific nature of the deviation to guide targeted remediation.

Does post-deployment drift detection require baseline reconstruction data?▼

Drift remediation involves analyzing current output patterns to classify drift signals, assess magnitude, and perform root cause analysis. This process generates actionable response recommendations to correct the AI system's shifted utility function.

What is the best way to monitor AI systems for constraint erosion and scope creep?▼

The best way to monitor AI systems for constraint erosion and scope creep is periodic drift detection comparing reconstructed baseline behavior against current output patterns. This classifies drift signals and measures magnitude for targeted response recommendations.

Related Skills

Explore More Data & Analytics Skills→
josemlopezjosemlopez
8

tm-drift

Compare current threat models against a baseline and generate drift-report.md.

Community
Intermediate
rory-datarory-data

debugging-dags

Diagnose failed Airflow DAGs and provide root-cause remediation.

Community
Advanced
IgorGanapolskyIgorGanapolsky
38

Position Sizer

Calculate optimal trade position sizes using ATR, Kelly Criterion, and volatility-adjusted methods.

Community
Advanced
pluginagentmarketplacepluginagentmarketplace
1

ml-monitoring

Detect data and model drift in production ML deployments.

Community
Advanced