dcorr

Detect and correct degraded signals and misjudgments in AI outputs.

4|Updated Feb 21, 2026
One-click install
npx skills add https://github.com/ssdiwu/diwu-workflow --skill dcorr
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dcorr
Source: https://github.com/ssdiwu/diwu-workflow/tree/main/skills/dcorr
Command: npx skills add https://github.com/ssdiwu/diwu-workflow --skill dcorr

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Skill provides a structured protocol to detect, stop, and correct degraded signals and misjudgments in AI outputs, reducing drift and improving reliability.

Core Features & Use Cases

  • Four-line rewrite template to anchor tasks to a single main line and current observations.
  • Drift detection and mitigation: predefined signals trigger a return to the main line and safer continuation.
  • Systematic misjudgment inspection to guide verification, edge-case handling, and robust AI behavior.

Quick Start

Simply trigger a correction cycle whenever outputs drift or misalign with the intended objective.

Frequently Asked Questions about dcorr

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

FAQPage Schema
How do I detect and correct drift in AI outputs across iterative prompts?

AI output drift is detected and corrected using a structured protocol that triggers predefined signals to return to the main task line and apply a four-line rewrite template for safer continuation.

What is the best way to prevent misjudgments in AI-assisted workflows?

Preventing AI misjudgments requires systematic inspection to guide verification and edge-case handling, applying a boundary-driven decision framework to distinguish valid corrections from blocked states.

How do I stop my AI task cycle from degrading signals and losing the main objective?

Stop signal degradation in task cycles by applying a four-line rewrite template that anchors tasks to a single main line and current observations, mitigating drift whenever predefined signals trigger.

When should I use a correction cycle versus a BLOCKED state for AI errors?

Use a correction cycle for fixable misjudgments and a BLOCKED state for boundary violations, distinguished by applying a structured boundary-driven decision framework during systematic inspection.

Does drift detection work without external dependencies?

Drift detection works without external dependencies by implementing a self-contained structured protocol that monitors predefined signals and applies a rewrite template to maintain output reliability.

Why does my AI model lose context and drift during long iterative tasks?

AI models lose context and drift during iterative tasks due to signal degradation, which is mitigated by anchoring tasks to current observations and triggering predefined signals to return to the main line.