ek-dev-add-signal

Integrate custom signals, calibrators, and decision policies into the ek framework.

Updated Jun 23, 2026
One-click install
npx skills add https://github.com/thorwhalen/ek --skill ek-dev-add-signal
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ek-dev-add-signal
Source: https://github.com/thorwhalen/ek/tree/main/skills/ek-dev-add-signal
Command: npx skills add https://github.com/thorwhalen/ek --skill ek-dev-add-signal

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill addresses the need to extend the reference-free quality estimation pipeline in the ek framework, allowing developers to integrate custom signals, calibrators, and decision policies for robust extraction validation.

Core Features & Use Cases

  • Custom Signal Integration: Add deterministic verifiers, intrinsic confidence signals, or agreement-based ROVER voting to the estimation pipeline.
  • Calibration & Decision Logic: Implement custom calibrators like Platt or Isotonic scaling and define selective policies such as Conformal or Cost-Sensitive gates.
  • Use Case: When building an OCR system, use this skill to add a custom regex-based verifier signal that flags extracted dates that do not match a valid calendar format before the decision policy accepts the output.

Quick Start

Register a new custom signal component by implementing the Signal protocol and adding it to the ek registry for use in the estimate_quality pipeline.

Frequently Asked Questions about ek-dev-add-signal

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

FAQPage Schema
How do I add custom validation signals to an OCR quality estimation pipeline?

To add custom validation signals to an OCR quality estimation pipeline, implement the Signal protocol and register your component within the ek registry system for integration into the estimate_quality workflow.

What is reference-free quality estimation for extraction tasks?

Reference-free quality estimation validates extraction and OCR outputs without requiring gold-standard references, using custom signals, calibrators, and decision policies to estimate data reliability.

Can I use Platt scaling and conformal prediction gates for OCR validation?

Yes, you can implement Platt or Isotonic scaling calibrators and define selective Conformal or Cost-Sensitive decision policy gates to control validation logic within the ek framework.

How do I add a regex verifier to flag invalid extracted dates in OCR?

You can add a regex-based verifier signal by implementing the Signal protocol to flag extracted dates that do not match a valid calendar format before the decision policy accepts the OCR output.

Do I need gold-standard references to calibrate extraction confidence scores?

No, you do not need gold-standard references to calibrate extraction confidence scores, because the ek framework specifically supports reference-free quality estimation using intrinsic confidence signals and custom calibrators.