Find and manage outputs

The Outputs panel collects the figures, tables, scripts, reports, and findings produced or saved in the current session.

Browse the output feed

Open Outputs in the right panel. The feed is ordered newest first and provides All, Figures, Tables, Scripts, Reports, and Evidence categories with item counts. Search filters visible deliverables by their available titles, tool names, summaries, or table headings. In Scripts, an extra filter can narrow the list by tool when more than one tool is present.

1. Choose a category or enter a search term. If an expected item is hidden, clear the query and select All.
2. Open the matching row. The artifact appears in the canvas with a title and the controls supported for its type.
3. Inspect or copy/download it, then use Outputs to return to the feed. For reports, open the document view and check which compiled formats are available before downloading.

• Figures: select a figure to open it full size. Scroll or use the zoom buttons to enlarge it, drag to pan, and double-click or choose Reset to center it again. Choose Fullscreen for a larger view, or download the PNG. When the run has code, select its tool label to open the matching script.
• Tables: select a table to open it in the canvas, then copy as Markdown or CSV. The feed's table action tooltip also lists LaTeX, but the current canvas exposes only Markdown and CSV copy buttons.
• Scripts: inspect the exact executed Python for a run. Open Run details to see recorded execution ID, seed, hashes, and edit ancestry when present. If Mizura saved a translation, switch to its language tab; translated code is marked Not run, because only the Python source was executed. Download the selected code or choose VS Code; this requires the desktop bridge and shows an error if opening fails.
• Reports: open the report document, preview compiled PDF when available, and download available formats. See Reports & exports.
• Evidence: revisit findings bookmarked from a response selection toolbar.

The count reflects feed items, not the number of analysis runs. Figure-producing script results are represented by their figures rather than duplicated in Scripts. In the canvas, use ←/↑ and →/↓ to move among items of the same type, or the on-screen arrows. Esc exits full screen or closes the open artifact. Use the Outputs button to return to the feed.

If a table row says it could not load, retry by selecting it again; if the item is still unavailable, see Troubleshooting. A VS Code launch error is shown in the code view; VS Code export buttons require the desktop bridge.

Export the session

Prerequisite: the session needs at least one script or figure. In Outputs, choose .ipynb, .qmd, or ZIP in the Session bar. The browser downloads the selected file; check its name and extension in your download list. If the bar is absent, no qualifying script or figure is currently in the session. These exports use saved execution records; see Choose and use an export for contents, replay limits, and data-sharing cautions.

Save and reuse evidence

Select text in an assistant answer to open its selection toolbar, then choose Evidence. A saved excerpt appears under Evidence in this session's Outputs and becomes available to @ mentions. If you do not see it, confirm the selection toolbar action completed and check the Evidence category. It is a saved excerpt for navigation and context, not an independent verification of the claim.

For sources cited during web research or added to the workspace, use the separate Sources panel. It can search, filter by papers or web results, sort, and download the currently filtered source list as Markdown.