How does the Table Filler prevent hallucinations?

The skill is governed by a strict guardrail that only allows it to pull data present in your evidence packs. It forbids the invention of facts and requires a citation for every single row.

Can I use this for long-form narrative inside table cells?

No, this skill specifically prevents long prose in cells. It is optimized for atomic facts and comparisons; narrative content should be moved to draft sections instead.

What happens if my evidence pack is missing data for the table?

The skill will identify this as an evidence gap. You should then enrich your evidence_drafts.jsonl file or adjust your schema rather than having the tool guess the information.

What files are required for this skill to function?

You need four primary inputs: a table schema (table_schema.md), subsection briefs, evidence drafts in JSONL format, and a BibTeX reference file (ref.bib).

Evidence-First Table Filler

Name: Evidence-First Table Filler
Author: WILLOSCAR

byWILLOSCAR

•

学习与文档

Populates structured tables with citation-backed facts directly from evidence packs to eliminate placeholders and prose.

The Evidence-First Table Filler is a specialized Claude Code skill designed for high-integrity technical writing and research documentation. By mapping table schema definitions to verified evidence drafts, it generates clean, professional Markdown tables that are strictly limited to atomic facts and mandatory citations. This skill solves the common issue of AI-generated 'hollow' writing by enforcing a methodology where every row must be backed by a source, effectively preventing hallucinations and the use of placeholder text like 'TODO' or ellipsis in your final documentation.

主要功能

01Zero-placeholder policy to ensure production-ready output

0289 GitHub stars

03Automated table generation from schema and evidence packs

04Anti-prose guardrails to keep table cells concise and readable

05Integration with research-units-pipeline evidence drafts

06Strict citation enforcement using BibTeX keys for every row

使用场景

01Populating data-driven technical documentation from verified research evidence

02Generating comparative technical feature tables in research papers

03Creating literature review summaries with direct source citations

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add willoscar/research-units-pipeline-skills table-filler

For use in Claude.ai and ChatGPT

Download Skill