Can this skill be used for any data pipeline?

No, this skill is specifically specialized for the contact-center-lab pipeline. It requires the specific directory structures and schema definitions found in that project to function correctly.

How does it determine if output quality is 'good'?

The skill derives expectations at runtime by reading your pipeline's source code and configuration. It then applies heuristics specific to knowledge base articles to check for appropriate entity counts, triple density, and graph connectivity.

What happens if my pipeline schema changes?

Because the skill 'derives, doesn't hardcode,' it automatically adapts to schema changes by inspecting your Python dataclasses and model definitions in real-time.

Does this skill modify my output files?

No, this is a strictly read-only skill. It analyzes your data and generates reports but never modifies files, commits, or pushes changes to your repository.

Pipeline Output Evaluator

Name: Pipeline Output Evaluator
Author: davistroy

bydavistroy

•

Analíticas y Monitorización

Performs deep semantic quality audits and structural validation for contact-center-lab pipeline outputs.

The Pipeline Output Evaluator is a specialized diagnostic tool designed to maintain high data integrity within the contact-center-lab ecosystem. It automates the complex task of validating multi-stage processing pipelines by dynamically discovering schemas from source code and comparing actual outputs against derived expectations. By tracing symptoms back to root causes and analyzing downstream cascades, it provides developers with actionable architectural insights rather than just surface-level error reports, ensuring that knowledge graphs and procedures remain accurate and sanitized.

Características Principales

01Automated detection of LLM failures and clustering bottlenecks

02Detailed root-cause analysis for every identified data quality finding

03Comprehensive semantic checks for entities, triples, and procedure integrity

04Regression analysis comparing current outputs against historical baselines

053 GitHub stars

06Dynamic schema and configuration discovery from pipeline source code

Casos de Uso

01Performing regression testing after updating pipeline logic or LLM models

02Verifying output quality after a full 10K corpus processing run

03Auditing PII sanitization and HTML artifact removal across datasets

Características Principales

01Automated detection of LLM failures and clustering bottlenecks

02Detailed root-cause analysis for every identified data quality finding

03Comprehensive semantic checks for entities, triples, and procedure integrity

04Regression analysis comparing current outputs against historical baselines

053 GitHub stars

06Dynamic schema and configuration discovery from pipeline source code

Casos de Uso

01Performing regression testing after updating pipeline logic or LLM models

02Verifying output quality after a full 10K corpus processing run

03Auditing PII sanitization and HTML artifact removal across datasets