knowledge-summarize

Generate a TL;DR summary by requesting it with a specific document_id or unit_id. The system fetches relevant indexed chunks, concatenates and truncates the content, and produces a concise structured markdown overview for rapid comprehension.

Summarize either a single PDF document by providing its document_id or an entire aggregated learning unit by providing its unit_id. Both options fetch indexed chunks to synthesize a quick overview with key points.

Document chunk retrieval fetches relevant text segments from pgvector-backed storage using the Knowledge API. It concatenates and truncates these chunks before passing them to the model to produce a structured markdown TL;DR.

The TL;DR output uses structured markdown featuring a concise summary and key points tailored for quick consumption. It includes a usage-oriented footer indicating the source chunks and documents used for rendering.

No manual chunk selection is needed to summarize a lesson. You simply provide the unit_id, and the system automatically retrieves, concatenates, and truncates the necessary chunks to generate the summary.

Summarizing large documents is subject to max token limits enforced during processing. The system truncates concatenated chunk content to fit these constraints before generating the final structured markdown summary.