Skip to main content
skill CX Codex

paper-context-resolver — Claude Code skill: finding dataset splits

paper-context-resolver looks up one specific missing detail in a paper's primary sources, such as a dataset split or eval protocol, when a repo's README can't answer it.

MIT 503,300 installs ★ 490 updated 2026-09-23

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add lllllllama/RigorPilot-Skills --skill paper-context-resolver -a claude-code
Open on GitHub

What paper-context-resolver does

Use it when a deep learning repo's README and files leave one reproduction question open: which dataset version or split, how inputs are preprocessed, how metrics are evaluated, which checkpoint matches which config, or what runtime is assumed. The skill pulls the answer from the paper's primary sources and labels what the paper states directly versus what is inferred.

The README stays the authority. If the paper disagrees with it, the skill records the conflict instead of overriding anything. You get a short source list and an answer to your question, not a paper summary.

It is an optional helper, usually called by the main orchestrating skill. It expects repo metadata, the concrete question, evidence already found in the README, and any known paper links. A bare paper title is not enough input: without a concrete question, this skill has nothing to resolve. Details sit in references/paper-assisted-reproduction.md.

What it helps with

  • find which dataset split the paper uses when the README doesn't say
  • check how the paper preprocesses data before training
  • pin down the evaluation protocol behind the reported metrics
  • match a released checkpoint to the paper's configuration
  • log where the README and the paper disagree

Example prompts for the agent

  • Use paper-context-resolver: the README doesn't say which dataset split to use, so check the paper's primary sources.
  • The README is silent on preprocessing. Ask paper-context-resolver to find it in the paper and flag any mismatch with the repo.
  • With paper-context-resolver, work out which checkpoint corresponds to the config described in the paper.

How to install paper-context-resolver

1Skills CLI (Vercel)

npx skills add lllllllama/RigorPilot-Skills --skill paper-context-resolver -a claude-code

2Manually: put the files in place

~/.claude/skills/paper-context-resolver/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/paper-context-resolver/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about paper-context-resolver

How do I install paper-context-resolver in Claude Code?

Run npx skills add lllllllama/RigorPilot-Skills --skill paper-context-resolver -a claude-code. The skill file ends up at ~/.claude/skills/paper-context-resolver/SKILL.md.

What does paper-context-resolver do?

It answers one narrow reproduction question for a repo, such as a dataset split or evaluation protocol, using the paper's primary sources. It keeps direct evidence separate from inference.

Does paper-context-resolver replace the README?

No. The README remains the main source, and the skill only fills a specific gap. When the paper contradicts the README, it writes the conflict down rather than silently overriding it.

Source text

SKILL.md · MIT
---
name: paper-context-resolver
description: Rigor Paper Context helper for README-first deep learning repo reproduction. Use only when the README and repository files leave a narrow reproduction-critical gap and the task is to resolve a specific paper detail such as dataset split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumption from primary paper sources while recording conflicts. Do not use for general paper summary, repo scanning, environment setup, command execution, title-only paper lookup, or replacing README guidance by default.
---

# paper-context-resolver

Use this as the Rigor Paper Context helper. The installed slug remains
`paper-context-resolver` for compatibility.

## When to apply

- README and repo files leave a reproduction-critical gap.
- The gap concerns dataset version, split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumptions.
- The main skill needs a narrow evidence supplement instead of a full paper summary.
- There is already a concrete reproduction question to answer.

## When not to apply

- The README already gives enough reproduction detail.
- The user wants a general paper explanation rather than reproduction support.
- The goal is to override README instructions without documenting the conflict.
- The only available input is a paper title and there is no concrete reproduction gap yet.

## Clear boundaries

- This skill is optional.
- This skill is helper-tier and should usually be orchestrator-invoked.
- It supplements README-first reproduction.
- It does not replace the main orchestration flow.
- It does not summarize the whole paper by default.

## Input expectations

- target repo metadata
- reproduction-critical question
- existing README or repo evidence
- any already known paper links

## Output expectations

- narrowed source list
- reproduction-relevant answer only
- explicit README-paper conflict note when applicable
- clear distinction between direct evidence and inference

## Notes

Use `references/paper-assisted-reproduction.md`.

© lllllllama · MIT · lllllllama/RigorPilot-Skills

Similar skills

All skills for Science and research →

From the same set11

The whole set →
minimal-run-and-audit minimal-run-and-audit is a Claude Code skill that captures and standardizes evidence from deep learning reproduction runs, including inference, evaluation, and smoke tests. CX Codex +1 502,400installs repo-intake-and-plan repo-intake-and-plan scans a repository through its README and project files, then builds a small reproduction plan. Use it before running experiments. CX Codex +1 502,400installs env-and-assets-bootstrap env-and-assets-bootstrap prepares environment and checkpoint/dataset paths before the first run of a README-documented repo. CX Codex +1 502,200installs ai-research-explore ai-research-explore governs deep learning research exploration on top of an existing baseline. It isolates candidate experiments and generates structured comparison reports. CX Codex +1 363,600installs analyze-project analyze-project reads a deep learning repository and maps its structure, entrypoints and suspicious spots. Changes nothing, runs no heavy jobs. CX Codex +1 363,500installs ai-research-reproduction ai-research-reproduction is a Claude Code skill for reproducing deep-learning runs from a repository README. It picks a small documented target and saves the evidence bundle to repro_outputs/. CX Codex +1 363,300installs safe-debug safe-debug is a Claude Code skill that diagnoses failures in deep learning research and drafts minimal patches without code changes unless approved. It activates when training fails. CX Codex +1 363,000installs explore-code explore-code — a skill for cautious code experiments in deep learning repositories: it transplants modules and adapts architectures on an isolated branch, with rollback-aware records. CX Codex +1 362,900installs

More in the catalog

Project sponsors

Companies that keep this analytics open