---
name: prior-use
description: Map a dataset's existing scientific uses and check the closest precedents for a proposed or revised research question, including conceptually equivalent questions under different terminology.
---

# Find what is known and what a question would add

Write all saved artifacts in English. Use the coordinator's assigned location; default to `research/<dataset>/knowledge/dataset/prior-use.md`. For older workspaces, link the current existing report rather than silently creating a competing version. Start from dataset identity, aliases, collection purpose and version; no seed is required. Use host search and source-reading tools, recording queries, dates, locators and consequential access limits.

1. Find collection/flagship and companion papers, then follow their citation/reuse trail and official publication lists for analyses of these observations. Include owner-reported published or ongoing work for private data; distinguish actual reuse from citations or related datasets.
2. Before formal question generation, prepare a readable account of the important uses, with a compact map as its index: evidence of actual reuse, version/subset, scientific question, comparison, findings and remaining uncertainty. For consequential papers write or link [detailed paper notes](../learn-patterns/references/paper-notes.md) so an unfamiliar reader understands what was done and why its inference follows. Mark abstract-only, inaccessible and unresolved sources. If no reuse was found, report searches and limits; state which parts of the literature remain unexamined.
3. For promising candidates, use [the precedent guide](references/precedent-hunt.md) to search conceptual equivalents and later work. Save the closest comparisons at the coordinator's assigned question directory or `precedents/<candidate>.md`, including the specific contribution left after reading them.
4. After a material reframe, update the affected searches and judge whether it is already answered, a substantive extension, a useful replication, or unresolved. Check subsequent work before treating an older paper's limitation as a current open question.

Report “not found within these searches” with scope and limitations rather than asserting novelty from absence. Follow decisive unresolved leads within budget; if an unread close source could reverse the judgment, carry that uncertainty into the candidate's disposition. Return the useful next search and the supported contribution to `discover`.