Changelog

0.2.0-beta.3 — installation prompt and listing assets

• Put a copyable installation request near the top of the homepage and in the quickstart, alongside direct installation commands.
• Explain that installation is available through GitHub independently of official directory listings.
• Register the existing blue mark as a 512-pixel PNG listing/composer icon and shorten the OpenAI listing subtitle to its 30-character limit.
• Provide a separate OpenAI upload ZIP containing the same plugin files without GitHub marketplace catalogs. The complete ZIP retains those catalogs for local installation.

The research workflow and expanded IBL example are unchanged. Earlier tags and archives remain fixed.

0.2.0-beta.2 — expanded scientific example

• Include the expanded IBL example: scientific motivation, closest studies, paired recording design, proposed computations, five possible outcomes and the checks that shaped the plan.
• Update the examples index and homepage to help readers find the scientific explanation.
• Synchronize the host manifests, citation version and installation guide at 0.2.0-beta.2 for directory submission.

The scientific workflow is unchanged. This release adds no new discovery or neural analysis; the IBL proposal remains a conditional example from alpha.3. The beta.1 tag and archive remain available unchanged.

0.2.0-beta.1 — research preview

• Refine the homepage, guides, examples and research-writing instructions for scientific readers; use a blue palette for the mark and plugin theme.
• Package the four-skill workflow with an installation guide, method explanation, result-reading guide, examples, support and privacy information.
• Use the stable maiwen marketplace identity; document migration from maiwen-local and preserve research workspaces.
• Include citation metadata and prominently credit Maieusis, the predecessor.
• Clarify that independent evaluation follows the affected scientific choices and claim; authorized planning exposure does not automatically require a new dataset.
• Correct the illustrative measurement example: equivalence of an overall mean protocol difference does not establish preservation of group contrasts.
• Keep the workflow from 0.2.0-alpha.5: independently reviewed field knowledge, reusable patterns and dataset knowledge; fresh question generation; separate planning and review; finite repair and one optional additional generation batch.

The validation record describes installation checks, document review and the earlier scientific development trials.