---
name: voltpeer-data
description: Prepare LAND/NEWARE exported CSV, TXT, XLSX and XLS battery tables, or import checked NEWARE NDAX NDC-14 source layers; identify declared units and retain original observations for plotting.
---

# Prepare battery test data

This Skill imports and prepares tables; it does not include the plot renderer. To draw the resulting quantitative Figure, separately install and invoke `$voltpeer-plot`. If that Skill is unavailable, deliver the prepared table and its actual readiness status, then explain the required plotting installation. Do not imply the single data ZIP can draw a Figure by itself.

<!-- execution-contract -->
For model/tool adaptation or resuming a task, read [the execution guide](references/EXECUTION.md). DeepSeek Flash uses short stages and checkpoints; DeepSeek Pro can plan larger text/evidence batches, with the same scientific checks. Reply in the user's language with the result, usable result/preview links and material unresolved questions. Keep mappings, configuration, logs and recovery records inside the project's `.voltpeer/` folder; do not link them in a normal final reply. Provide the corresponding source record only when the user explicitly requests provenance. Use only capabilities actually available in the current model and host.
<!-- /execution-contract -->

Use this before plotting when the author brings raw cycling, impedance, spectroscopy or other instrument tables. Start by inventorying files and reading a few rows; an extension or instrument filename does not identify the experiment, voltage reference, capacity basis or units. Keep the raw export untouched in the author's workspace. Do not upload unpublished data to a public repository or issue.

For LAND / 蓝电, NEWARE / 新威 or a workbook with cycle, step and record layers, read [instrument exports](references/INSTRUMENT_EXPORTS.md) and run `scripts/instrument_import.py inspect`, then `prepare` into a new project directory. Read every sheet; the assistant selects and maps the actual table, so the user need not rename columns or edit JSON. The helper retains original cells and adds separately labelled canonical fields using declared units. Checked NEWARE NDAX NDC-14 files retain measurement, step and run-information layers separately; duplicated metadata or unresolved alignment cannot become a guessed curve. CEX, untested NDA and unsupported NDAX use the stated vendor-export route. Resolve only units, cell/branch meaning, normalisation and conditions needed for the requested plot.

For a plain CSV, run `python scripts/prepare_csv.py inspect --input RAW.csv`. It reports headers, row count, blanks and whether values are parseable as numbers; it does **not** assign scientific meanings. Ask the author to confirm the mapping, units and protocol where they are ambiguous. Then make a mapping JSON following [the exact schema and example](references/MAPPING_CONTRACT.md) and run `python scripts/prepare_csv.py prepare --input RAW.csv --mapping MAP.json --output-dir NEW_DIRECTORY`. The helper refuses an existing output directory, preserves row order and cell values, and writes `mapped.csv` plus a machine-readable record of hashes, mapping, units and every explicitly excluded row. It never converts units or drops a bad value on its own. For Excel or proprietary formats, inspect/export with a suitable available reader and document the export step before using this CSV helper.

Treat preparation as ready for quantitative plotting only after the author confirms cell configuration, branch/step identity, time or cycle basis, sign convention, normalization denominator and relevant test conditions. Record that review in `plot_context` and set `plot_context_confirmed: true` only after the author checks it; this is required for the helper's `ready_for_plot` flag. The flag records a human decision and does not prove the conditions correct. Unknown units remain `unknown` in the record and also block readiness. Do not guess units from magnitudes or convert mA to mA cm⁻² without measured area. Keep zero, missing, instrument flags and rejected points distinct. An exclusion requires a specific row number and reason supplied by the author; report how many rows remain and show exclusions for approval.

Show the prepared table link, declared units and any material unresolved question or exclusion count. Keep raw paths/hashes, mappings, detailed exclusions and checks in the private project record; expand them when the author asks or when needed to explain a decision. Route plotting to `voltpeer-plot`; route cross-study metric comparisons to `voltpeer-metrics`. This skill prepares evidence, not a scientific interpretation.