What problem does it solve?
Centralizes maintenance of Jinja2 prompt templates by detecting missing headers, broken includes, orphaned templates, encoding issues, and include cycles so teams can avoid runtime errors and undocumented prompts.
Core Features & Use Cases
- Comprehensive audit: Scan configured template sources, parse header metadata, map {% include %} relationships, and report structured errors and warnings.
- Python-aware orphan detection: Scan Python code (PromptLoader, get_template, PackageLoader patterns) to find direct template references and avoid false orphan reports.
- Catalog and sync: Generate per-source prompts.json, human-readable catalogs, and a combined catalog with a Mermaid dependency graph for documentation and CI integration.
- Single-file validation: Validate individual templates for header format, context/output fields, encoding, and include existence for pre-commit hooks.
Quick Start
Run the prompt-maintenance audit to scan all template sources, map dependencies, and produce an errors-and-warnings report.