agentsframework-axial-coding

To perform axial coding on open-coded qualitative data, you automate the categorization and clustering of codes into named, testable failure categories using Python scripts and CSV files.

Axial coding in grounded theory is the Stage 2 process of categorizing and clustering open-coded data into structured, named failure categories to generate testable categories and enable further analysis.

You create a failure taxonomy from open-coded data by running an automated axial coding pass on a coded JSONL file, which clusters open codes into testable categories and generates rubric assertions.

Yes, you can use axial coding for AI agent evaluation by processing open-coded qualitative data to identify minimal pairs and generate judge test-case candidates for structured failure analysis.

Automated axial coding requires a JSONL file containing open-coded data, which the Python scripts process and organize to output structured failure categories and testable rubric assertions.