Does this skill help the AI avoid making the same mistake twice?

Yes. By documenting the 'before' and 'after' state of an analysis, future tasks can reference these logs during validation steps to ensure previous errors aren't repeated.

Where are the correction logs stored?

Corrections are saved locally in your workspace under the path ` /knowledge/corrections/` in YAML format.

Can I log corrections if there is no SQL involved?

Absolutely. While it handles SQL well, it is also designed to capture flawed reasoning, metric definitions, and general logic errors.

How do I trigger the Log Correction skill?

You can trigger it by using the command `/log-correction` or by telling Claude 'log a correction' or 'that was wrong because' during a conversation.

What types of errors can I log?

The skill categorizes errors into SQL query issues, metric definitions, schema mismatches, flawed logic, and other general analytical mistakes.

AI Analyst Log Correction

Name: AI Analyst Log Correction
Author: ai-analyst-lab

byai-analyst-lab

•

Analíticas y Monitorización

Records and categorizes analytical mistakes to build a persistent knowledge base that prevents recurring errors in future analyses.

The Log Correction skill provides a structured framework for Claude to learn from its mistakes by manually capturing feedback, errors, and their corresponding fixes. When an analytical error is identified—such as a flawed SQL query, a misunderstood metric definition, or a logic gap—this skill guides the user through documenting the issue, assigning a severity level, and updating a local knowledge repository. By maintaining a historical log of corrections, the AI can perform self-validation in future tasks, ensuring that past failures become the foundation for higher accuracy and more reliable data insights.

Características Principales

01Automated categorization by error type including SQL, metric, and schema

02Manual feedback capture via natural language or command-line triggers

0319 GitHub stars

04Severity-based prioritization for critical data accuracy tracking

05Structured YAML-based logging of SQL snippets and logic corrections

06Continuous index updates to facilitate future validation and learning

Casos de Uso

01Refining AI-generated SQL queries after discovering a join or filter error

02Documenting specific business logic nuances that the AI initially missed

03Building a team-wide knowledge base of common data pitfalls and their resolutions

Características Principales

01Automated categorization by error type including SQL, metric, and schema

02Manual feedback capture via natural language or command-line triggers

0319 GitHub stars

04Severity-based prioritization for critical data accuracy tracking

05Structured YAML-based logging of SQL snippets and logic corrections

06Continuous index updates to facilitate future validation and learning

Casos de Uso

01Refining AI-generated SQL queries after discovering a join or filter error

02Documenting specific business logic nuances that the AI initially missed

03Building a team-wide knowledge base of common data pitfalls and their resolutions