Choose and use an export

Choose the file for the next step: sharing a written result, editing code, or preserving the recorded runs for replay.

Pick a format

If you need to…	Choose	What it contains
Share or submit the written findings	Report	Checked report text, tables, figures, and citations in a document format.
Inspect or continue the recorded Python work	Jupyter notebook (`.ipynb`)	The plan and decisions followed by one code cell per kept execution, with recorded standard output.
Use a literate analysis document	Quarto (`.qmd`)	Plan and execution sections with Python chunks disabled by default; tables and figures link to session files. Download those artifacts separately and update paths before rendering the QMD outside Mizura.
Keep records and replay executions	Replication package (`.zip`)	README, report when available, execution code and records, recorded inputs, tables, figures, and any code translations.

Before you export: finish the run you want to keep and check its outputs. Exploration scratch runs are omitted. The session export controls appear after the session has a script or figure; they do not export the report document itself.

For a formatted report, start with Reports & exports. It covers report formats, editing, table and figure downloads, and report checks.

Download a session export

Open the session you want to keep.
Open Outputs.
In the session export bar, choose .ipynb, .qmd, or ZIP. The bar appears after the session has a script or figure.
Save the downloaded file where you want to keep it.

Notebook, Quarto, and replication exports are built from the engine's saved execution records. They keep the latest successful execution for each analysis; exploratory scratch runs are left out. A Quarto file does not run its code chunks automatically. Review code before running it in another environment.

Success check: confirm the downloaded file has the selected extension. Open a notebook to review its cells, a Quarto file to review its disabled code chunks and linked artifacts, or the ZIP to check its README and execution folders. If the export controls are missing, complete an analysis that produces a script or figure. If a downloaded file says there is no analysis to export, the session may contain only exploratory scratch runs. If a download fails or the file will not open, retry after the run has finished and inspect the session outputs before relying on it.

Know what the package can replay

The replication package includes each kept execution's code, execution record and hashes, captured output, tables, figures, and the files staged as inputs. Use the README inside the ZIP to find the replay command and the execution IDs. A replay marked matchedmeans the run reproduced the recorded status and hashes under the current runtime.

Unzip the package and open its README. Copy the execution ID you want to replay from its table.
With misura-core installed in the Python environment, run the README command from the unzipped package directory, replacing EXECUTION_ID with that ID. For a large-data run, follow the note below and rebuild its table first.
Check the returned matched value. true means the recorded status and hashes match; false means this run did not reproduce the record.

python -c "import os; from misura_core.sandbox.replay import replay_execution as r; print(r('EXECUTION_ID', session_output_dir=os.getcwd()))"

Large datasets are a special case: the package omits the materialized dataset.duckdb table to avoid putting a potentially huge file in the ZIP. To replay those runs, keep the original dataset and rebuild the table; the README notes this requirement and records the hash of the omitted table. Compare the recorded hash; matching input files alone do not guarantee a match if the software environment has changed.

If replay reports a mismatch, compare the execution ID, input files, status, and recorded hashes in the README and execution record. Keep the original dataset for large-data runs. A mismatch means this replay did not match the record; it does not identify which cause changed.

Before sharing. Open the package and inspect its files. It can include data supplied as run inputs, session report text, and outputs. Remove or protect sensitive data according to your research agreement. A downloaded export remains after you delete its session.

For individual outputs, including figures, CSV tables, scripts, and datasets, see Per-item exports.