Inspect the source passage
Open Papers & sources to work with extracted paper content. Follow a reference to the document and passage it points to, then read enough surrounding context to judge whether it supports the claim.
Extraction can lose tables, equations, or layout. A referenced passage is a starting point for review, not an automatic endorsement of the assistant’s interpretation. Reopen the preserved original when the wording or context is unclear.
Separate findings from claims
A calculation result, an interpretation of that result, and a statement from a paper are different kinds of evidence. Inspect the saved artifacts and claim references to understand which one supports an answer.
When the evidence is incomplete, retain that limitation. A narrow supported conclusion is more useful to the next investigation than an overconfident answer with a citation attached.
Record unsuccessful attempts
Ask your research group to preserve what was tried, its outcome, the relevant inputs, and what you learned. Distinguish a failed calculation from a valid negative finding or a scientifically invalid comparison.
Before reusing an earlier lesson, check that its conditions match the new investigation. Saved context helps avoid repeating work; it does not make conclusions transferable to unrelated datasets.
Export a reproducible record
Research runs retain their selected materials, results, and provenance in reports. Open the report linked from the research result to inspect and download the archive.
The report captures a fixed snapshot. Later project changes do not alter that archive. Keep the archive with the study and use the repository’s report validation and replay guide when you need to inspect hashes, compatibility, or rerun supported scientific steps.