carta-cap-table:6.91.7

Excel Export

Context expected from the calling skill (must be in session before this skill is invoked):

• user_report_pk — needed to check for the cached report file and to fetch a fresh download URL if the file is absent
• corporation_id — needed for call_tool({"name": "reporting__get__download_url", ...})
• Column config — either parsed from the artifact prompt bar payload (Claude Desktop) or confirmed during the Customization Checkpoint in carta-reporting-markdown (Claude Code)
• Corporation legal name, as_of_date, user full name — used as --title, --as-of-date, --generated-by args to excel_exporter.py

Column source (Claude Desktop — artifact prompt bar)

When the user pastes an artifact prompt bar payload, it arrives as a complete instruction starting with Generate Carta Excel —:

Generate Carta Excel —
Corporation: Meetly, Inc. (ID: 7)
Columns:
  Equity Grants: columns: Grant ID, Award Type, Exercise Price; sorted by: Grant Date desc
  Vesting Schedule: columns: Grant ID, Vest Date, Shares Vested; totals: Shares Vested sum

Use the Corporation: line to confirm which company the export is for. Each indented line under Columns: is one sheet tab. This is always one Excel file with one tab per sheet. Parse each sheet line and pass all sheets to report_processor.py in a single run using the per-sheet sheets dict format. No further questions needed.

The column list from the artifact is authoritative; do not merge with or override it from the earlier conversation.

Parsing segment fields into report_processor.py config:

totals: must become aggregations — this is the only way Excel formulas (=SUM(...), =AVERAGE(...)) are generated. If aggregations is omitted, any total rows in the output are hardcoded API values, not live formulas.

If the user asks for Excel without pasting a payload, ask: "Click the Excel export bar at the bottom of the artifact to select it, copy and paste it here — I'll generate the Excel with exactly those columns."

No scope context on this path. Entering fresh from the prompt bar, this skill cannot know whether the account is share-class-scoped, so it cannot label a percentage column for the scope or name the in-scope classes — the same limitation as label_overrides. The numbers are correct and already scoped; only the labelling is generic. Do not describe any total as company-wide.

Column source (Claude Code)

Use the column list confirmed during the Customization Checkpoint (resolved in carta-reporting-markdown).

Running the export

Check if /tmp/carta_report_<user_report_pk>.json is available (use user_report_pk from this session):

• File ready → pass it as "local_file" to report_processor.py.
• Not ready → call call_tool({"name": "reporting__get__download_url", "arguments": { user_report_pk, corporation_id }}) to get a fresh presigned URL and pass it as "download_url" instead.

Always pipe through report_processor.py → excel_exporter.py, regardless of data size or complexity. Never write Excel files directly with openpyxl or any other library — the scripts handle Carta branding (logo, header, fonts, number formats) that will be missing from any ad-hoc implementation. This applies even when the dataset is small (e.g. 3 rows) or when sheets need to be combined.

For combining sheets into one tab, use merge_sheets in the report_processor.py call.

Pass all sheets in one run using the sheets dict. Pipe into excel_exporter.py. Reuse the cached _report_processor_path if the parent session resolved it (it may be unset on the fresh Generate Carta Excel — prompt-bar path — the find fallback handles that):

UV_PYTHON_DOWNLOADS=never uv run "${_report_processor_path:-$(find ~ -name "report_processor.py" -path "*/carta-reporting/scripts/*" 2>/dev/null | head -1)}" <<'EOF' | \
    UV_PYTHON_DOWNLOADS=never uv run "$(find ~ -name "excel_exporter.py" -path "*/carta-reporting-excel/scripts/*" 2>/dev/null | head -1)" \
        --title "Securities Ledger Report" \
        --as-of-date 2024-01-15 \
        --generated-by "Jane Doe" \
        --output ./{report-slug}.xlsx
{
  "local_file": "<path or use download_url if file not ready>",
  "sheets": {
    "Equity Grants":    {"columns": ["Grant ID", "Award Type", "Exercise Price"],
                         "aggregations": {"type": "summary", "columns": {"Exercise Price": "sum"}}},
    "Vesting Schedule": {"columns": ["Grant ID", "Vest Date", "Shares Vested"],
                         "sort": [{"column": "Vest Date", "direction": "asc"}]}
  }
}
EOF

The script prints the absolute output path on success. Present it as a clickable link: computer://<absolute-path> (e.g. computer:///Users/jane/meetly-equity-grants.xlsx). Tell the user their file is ready to open, then offer next steps:

• Run another report — for a different company or report type
• Customize this export — adjust filters, columns, or formulas
• Change the date range or filters — re-run with different parameters

Carta Excel Formatting Conventions

Column type → number format: money → $#,##0.00 · percentage → 0.00% · integer → #,##0 · date → mmm d, yyyy · decimal → #,##0.0000

Column widths: string/date → 35, number types → 18.