table-filler_skill

Overview

This skill fills outline/tables_index.md from a table schema plus evidence packs to produce compact, citation-backed index tables. It creates planning/traceability tables (not paper-ready prose) that surface representative works and anchor facts with explicit citations. The output is deterministic and strictly evidence-first.

How this skill works

It reads outline/table_schema.md to learn required table axes, then pulls subsection briefs, evidence drafts, anchor sheet entries, and the bibliography to populate rows. Cells are short, non-prose entries and every row includes one or more citations in [@BibKey] format. The skill validates cite keys against citations/ref.bib and flags gaps rather than inventing facts.

When to use it

• You have a finalized table schema and assembled evidence packs
• You need a compact, citation-backed index to guide writing or appendix curation
• You want to check coverage and traceability before drafting manuscript sections
• You need anchor facts (benchmarks, numbers, caveats) mapped to sources
• You want deterministic, reproducible table-generation from evidence

Best practices

• Run only when outline/tables_index.md is missing or not refined (fewer than 2 tables or missing citations)
• Ensure outline/subsection_briefs.jsonl and outline/evidence_drafts.jsonl are up-to-date before running
• Keep cells short: phrases, keywords, or numeric entries; avoid sentences
• Treat any unfillable cell as an evidence gap and route to evidence drafting workflows
• Validate citations/ref.bib is current so all cite keys resolve

Example use cases

• Generate a subsection map table listing research axes with representative citations from evidence drafts
• Produce an anchor-fact table summarizing benchmarks, numeric results, and caveats with source keys
• Rapidly audit coverage: identify schema fields lacking citation-backed evidence
• Create a deterministic index artifact to hand off to the appendix-table-writer for reader-facing tables