RAG eval harness starter · Python + Markdown
The eval framework scaffold we drop into RAG projects in week one. Giskard + promptfoo + custom metrics.
Starter scaffold for RAG system evaluation: 5 metric classes (faithfulness, context precision, answer relevance, bias, injection resistance), CI integration, diff reporting. MIT-licensed.
WHAT'S INSIDE
- ·5 metric classes with Python code
- ·Giskard + promptfoo integration
- ·Injection-resistance eval suite (80+ prompts)
- ·CI step YAML (GitHub Actions + GitLab CI)
- ·Diff-report generator per build
Want a custom version?
A tailored audit or template delivered in 2 weeks · DField Solutions, Budapest.