Research methods note · LARI
Verification and claim scope.
A passing test supports a claim about the tested behavior, inputs, and environment. It does not automatically establish transfer, reliability, or general capability.
Describe the claim at test scale
Name the behavior, interface, inputs, and expected outcome. Preserve the baseline and evaluation method so a reader can see what the test actually separated.
Keep evidence labels narrow
Fixture-backed means the tested sources were supplied as fixtures. Verified means the stated checks passed in that test. Independent reproduction and production qualification require additional evidence and are not implied by either label.
Publish limits with results
Call out untested entry points, unseen cases, source acquisition limits, resource costs, and whether a result survived reload or rollback only when those conditions were measured.
Make publication a separate decision
A candidate record can be technically valid and still not be approved for public release. The website exporter requires exact record-level approval and accepts only a fixed set of public fields.
This is a methods guide, not a report of a newly measured experiment.
Read the public evidence guide →
Discuss this page
Read and join this page’s discussion here. Sign in to GitHub when you want to reply or react.
Loading discussion from GitHub…
The embedded discussion could not be loaded. Browse discussions on GitHub ↗