check-sentence-by-sentence

Audit claim-bearing sentences in an ASTRA paper against code locations.

18|2|Updated Feb 15, 2026
One-click install
npx skills add https://github.com/LightconeResearch/lightcone-cli --skill check-sentence-by-sentence
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: check-sentence-by-sentence
Source: https://github.com/LightconeResearch/lightcone-cli/tree/main/claude/lightcone/skills/check-sentence-by-sentence
Command: npx skills add https://github.com/LightconeResearch/lightcone-cli --skill check-sentence-by-sentence

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill finds exactly which implementation or results claims in a paper are supported by the code in an ASTRA project, and which ones are missing or not traceable.

Core Features & Use Cases

  • Sentence-by-sentence static audit: For each claim-bearing sentence, it locates the corresponding code location (file:line) or marks it NOT FOUND.
  • ASTRA-spec guided mapping: It uses astra.yaml narrative.method and narrative.findings (plus prior_insights/evidence where present) to translate paper prose into decision/output IDs and then into code.
  • lc-from-paper support: It reads the paper substrate from work/reference/ (prefer TeX under work/reference/source/, otherwise work/reference/document.md) without requiring any code execution.

Quick Start

Ask an AI agent to run check-sentence-by-sentence from the project folder that contains astra.yaml and point it at your paper source, for example work/reference/source/main.tex.

Frequently Asked Questions about check-sentence-by-sentence

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify paper claims against my codebase without executing code?

Static analysis of paper claims against code verifies methodology and results by mapping claim-bearing sentences to specific file:line locations in your project, returning a per-sentence JSON list with quote-to-code mappings or NOT FOUND.

How do I trace paper prose to specific code locations in an ASTRA project?

Tracing paper prose to code uses the astra.yaml narrative.method and narrative.findings fields to translate text into decision and output IDs, locating corresponding file:line references across methodology, results, discussion, and appendix sections.

Can I audit a LaTeX paper to check if its claims are supported by implementation?

Yes, auditing a LaTeX paper is supported by reading the TeX source from work/reference/source/ (or work/reference/document.md), enumerating sections by headings, and statically mapping claim-bearing sentences to code locations without running anything.

What is static reproducibility analysis for research papers?

Static reproducibility analysis is the process of auditing claim-bearing sentences in a paper against implementation and result computations present in a codebase, ensuring methodology traceability and code provenance without executing the project.

Does paper verification require a specific project file structure to work?

Paper verification requires locating the project root astra.yaml file to guide the mapping process, and expects the paper substrate to exist in work/reference/ as either TeX source files or a markdown document.

What happens to paper claims that cannot be found in the codebase?

Claims that cannot be traced to specific implementation or result computations in the codebase are explicitly marked as NOT FOUND in the per-sentence JSON list, highlighting missing or untraceable methodology evidence.