documentador-rag

What problem does it solve?

Converts heterogeneous, unstructured inputs into deterministic, canonical Markdown documents explicitly prepared for vector-database ingestion and semantic chunking. It enforces fidelity to the original input, prevents undocumented summarization, and structures content so downstream pipelines can reliably separate theoretical and practical chunks.

Core Features & Use Cases

• Deterministic Markdown output: Produces canonical .md files with a required YAML metadata block and rigid sectioning to support automated chunking.
• Theory vs Practice separation: Marks content structurally as Teoria and Pratica so each chunk's role is explicit for retrieval and pipeline processing.
• Broad input support: Designed for transcripts, meeting notes, Swagger/API specs, raw JSON, and technical documents to build knowledge bases and RAG corpora.
• Quality rules: Enforces no summarization, per-section Summaries optimized for embeddings, and preservation of all technical details present in the input.

Quick Start

Use the documentador-rag skill to convert meeting-audio.mp3 into a deterministic Markdown document with required YAML metadata, separated theory and practice sections, and Summary blocks suitable for semantic chunking.