| name | enrich |
| description | Enrich paper metadata using LLM extraction. Extract table of contents (TOC), generate L3 paper-level conclusion cards, backfill abstracts, or refetch citation counts from APIs. Use when the user wants to build TOC, update L3, update citation data, or backfill missing abstracts. |
Enrich Paper Content
Use LLM to extract a table of contents (TOC) or generate an L3 paper-level conclusion card. L3 first tries to locate an explicit conclusion/summary section; if none exists, it synthesizes a constrained takeaway from abstract, results, discussion, table/caption text present in Markdown, and highlights. MinerU image attachments are not retained.
Note: import-endnote / import-zotero automatically run toc + l3 + abstract backfill by default. The commands below are for selective enrichment — re-extracting specific papers, supplementing individual entries, or processing the entire library.
Execution Logic
-
Determine the user's intent:
- Extract TOC: use
enrich-toc
- Generate or refresh L3: use
enrich-l3
- Backfill abstracts: use
backfill-abstract (extracts from .md + LLM validation)
- Refetch citation counts: use
refetch (re-queries APIs to fill in citation_count and related fields)
-
Determine the scope:
- Specify a paper ID → process that single paper
- User says "all" → use
--all
- Optionally add
--force to overwrite existing results
-
Run the command:
Extract TOC:
autor enrich-toc [<paper-id> | --all] [--force]
Generate L3:
autor enrich-l3 [<paper-id> | --all] [--force]
Backfill abstracts:
autor backfill-abstract [--dry-run]
Refetch citation counts:
autor refetch [<paper-id> | --all] [--force]
- Display the processing results.
Examples
User says: "Extract the conclusions for all papers."
→ Run enrich-l3 --all
User says: "Re-extract the TOC for Smith-2023-Survey."
→ Run enrich-toc "Smith-2023-Survey" --force
User says: "Backfill missing abstracts."
→ Run backfill-abstract, then index --rebuild
User says: "Refresh the citation counts."
→ Run refetch --all, then index --rebuild