🚀Shippingscore 93.3May 15, 2026·2605.16194cs.DLcs.AIcs.IRcs.MA

paper.json: A Coordination Convention for LLM-Agent-Actionable Papers

Arquimedes Canedo

Narrative

A companion JSON schema for academic papers that gives LLM agents stable, citable anchors for sub-claims (C1), an explicit list of what the paper does *not* claim (C2), exact shell commands to reproduce each figure (C3), and stable definition IDs (C5). The core bet is that current prose PDFs cause systematic agent failures — scope hallucination, non-reproducible figures, un-citable sub-claims — and a lightweight sidecar file costing under an hour to write fixes all of them without changing the PDF itself.

Zero citations and no production traction yet. The referenced repos are all arXiv digest aggregators, not implementations. The paper's own validator (`uv run validator.py paper.json --against paper.typ`) exists at the author's GitHub, but adoption depends on whether authors voluntarily ship `paper.json` alongside PDFs — a coordination problem the convention itself can't solve unilaterally.

Abstract

LLM agents routinely serve as first (and sometimes only) readers of academic papers, skimming for sub-claims, extracting reproducibility steps, and generalizing scope. Standard prose papers produce recurring failures in this role: sub-claims that cannot be cited at sub-paper granularity, scope overextension beyond what the paper tests, and figure commands buried in codebases rather than the paper itself. We propose `paper.json`, a companion JSON file that travels with the PDF and addresses each failure with a lightweight convention: stable claim IDs (C1), an explicit does-not-claim list (C2), exact per-figure shell commands (C3), and stable definition IDs (C5). A fifth convention (C4) holds that minimum viable compliance, hand-written JSON alongside the PDF, is achievable in under an hour for a finished paper without touching the human-readable output. C1, C2, C3, and C5 are open invitations: an agent that reads a compliant paper and acts on it produces evidence for or against them. This paper is itself compliant: `uv run validator.py paper.json --against paper.typ` passes. Repo: https://github.com/arquicanedo/paper-json

Citation timeline

Not enough citation snapshots yet to plot a timeline. Come back after a few cron runs.

Signal

Stars: 345
Repos: 19
Citations: 0
Velocity: 0.00/d

GitHub repos (19)

luohongk/Embodied-AI-Daily⭐ 245
“| **[Confirming Correct, Missing the Rest: LLM Tutoring Agents Struggle Where Feedback Matters Most](https://arxiv.org/abs/2605.16207v1)** | 2026-05-15 | 22 pages, 20 fgures | | **[Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adve”
CSQianDong/Awesome-arXiv-Daily-Reporter⭐ 47
“{'arxiv_id': 'arXiv:2605.15213', 'title': 'An LLM-RAG Approach for Healthy Eating Index-Informed Personalized Food Recommendations', 'authors': 'Yibin Wang, Yanjie Yang, Grace Melo Guerrero, Rodolfo M. Nayga Jr., Azlan Zahid', 'link': 'https://arxiv.org/abs/2605.15213', 'abstract”
ehijano/rss_fetch⭐ 11
“ </item> <item> <title>paper.json: A Coordination Convention for LLM-Agent-Actionable Papers</title> <link>https://arxiv.org/abs/2605.16194</link> <description>arXiv:2605.16194v1 Announce Type: cross Abstract: LLM agents routinely serve as first (and som”
ZenAlexa/agi-brief-history⭐ 11
“- **Summary**: We examine one particular dimension of AI governance: how to monitor and audit AI-enabled products and services throughout the AI development lifecycle, from pre-deployment testing to post-deployment auditing. Combining principles from formal methods with SoTA mach”
lonePatient/lonePatient.github.io⭐ 9
“{% hideToggle 点击查看摘要 %} {% note blue no-icon %} ID-29-paper.json: A Coordination Convention for LLM-Agent-Actionable Papers {% endnote %} **链接**: https://arxiv.org/abs/2605.16194 **作者**: Arquimedes Canedo **类目**: Digital Libraries (cs.DL); Artificial Intelligence (cs.AI); Inform”
jyyang621/DailyArXiv⭐ 8
“| **[Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDP](https://arxiv.org/abs/2605.16205v1)** | 2026-05-15 | | | **[paper.json: A Coordination Convention for LLM-Agent-Actionable Papers](https://arxiv.org/abs/2605.16”
Guesswhat-Studio/Linnet⭐ 8
“ "abstract": "This paper introduces `paper.json`, a companion JSON file to academic PDFs, designed to improve LLM agent comprehension. Its core method is a set of lightweight conventions for stable claim IDs, explicit \"does-not-claim\" lists, per-figure shell commands, ”
2shin0/arxiv-ai-mailing⭐ 6
“ ## 37. paper.json: A Coordination Convention for LLM-Agent-Actionable Papers - **Authors**: Arquimedes Canedo - **URL**: [https://arxiv.org/abs/2605.16194](https://arxiv.org/abs/2605.16194) - **Abstract**: > LLM agents routinely serve as first (and sometimes only) readers of aca”
shaijing/arxiv-paper⭐ 0
“| **[Confirming Correct, Missing the Rest: LLM Tutoring Agents Struggle Where Feedback Matters Most](https://arxiv.org/abs/2605.16207v1)** | 2026-05-15 | <details><summary>Show</summary><p>Effective tutoring requires distinguishing optimal, valid but suboptimal, and incorrect stu”
sirichen2/sirichen2.github.io⭐ 0
“ <span class="paper-flag">Matched to your radar</span> <h2><a href="https://arxiv.org/abs/2605.16194v1" target="_blank" rel="noopener noreferrer">paper.json: A Coordination Convention for LLM-Agent-Actionable Papers</a></h2> ”
yzbcs/yzbcs.github.io⭐ 0
“ <div class="card-title"> <a href="http://arxiv.org/abs/2605.16194v1" target="_blank">paper.json: A Coordination Convention for LLM-Agent-Actionable Papers</a> </div>”
zhangluustb/awesome_multimodal_search_arxiv_daily⭐ 0
“| An LLM-RAG Approach for Healthy Eating Index-Informed Personalized Food Recommendations | [https://arxiv.org/abs/2605.15213](https://arxiv.org/abs/2605.15213) | Diet quality is a leading determinant of chronic disease risk. Advances in artificial intelligence (AI) have enabled ”
BenAmpel/Bampel_Website⭐ 0
“ "title": "paper.json: A Coordination Convention for LLM-Agent-Actionable Papers", "summary": "LLM agents routinely serve as first (and sometimes only) readers of academic papers, skimming for sub-claims, extracting reproducibility steps, and generalizing scope. Standa”
Jinsu-L/DailyIR⭐ 0
“ { "title": "paper.json: A Coordination Convention for LLM-Agent-Actionable Papers", "abstract": "LLM agents routinely serve as first (and sometimes only) readers of academic papers, skimming for sub-claims, extracting reproducibility steps, and generalizing sc”
lxl-sword/arxiv_paper_LLM_list⭐ 0
“<div class='col-md-6 mb-4'> <div class='card'> <div class='card-body'> <h4 class='card-title'><a href='https://arxiv.org/abs/2605.16194v1' target='_blank'>paper.json: A Coordination Convention for LLM-Agent-Actionable Papers</a></h5> <h5 class='card-subtitle mb-2 text-muted'>Auth”
mghnasiri/PORID⭐ 0
“ </item> <item> <title>paper.json: A Coordination Convention for LLM-Agent-Actionable Papers</title> <link>http://arxiv.org/abs/2605.16194v1</link> <guid isPermaLink="true">http://arxiv.org/abs/2605.16194v1</guid> <description>Arquimedes Canedo ”
mickdur/tech-watch⭐ 0
“ "https://arxiv.org/abs/2605.16143": "2026-05-18T07:51:44.206446+00:00", "https://arxiv.org/abs/2605.16153": "2026-05-18T07:51:44.206446+00:00", "https://arxiv.org/abs/2605.16165": "2026-05-18T07:51:44.206446+00:00", "https://arxiv.org/abs/2605.16194": "2026-05-18T07:51:44”
mirae0708/steven⭐ 0
“ > **Source:** [arXiv](http://arxiv.org/abs/2605.16194v1) > **Category:** Artificial_Intelligence/LLM”
pchaganti/pchaganti.github.io⭐ 0
“ { "title": "paper.json: A Coordination Convention for LLM-Agent-Actionable Papers", "summary": "LLM agents routinely serve as first (and sometimes only) readers of academic papers, skimming for sub-claims, extracting reproducibility steps, and generalizing scope. ”