Headless Excel Workbook Builder
Generate deliverable .xlsx files in managed-agent sessions without an open Excel application.
The skill limits output to the local ./out/<name>.xlsx path, does not request network access or external writes, and says not to overwrite an existing workbook, reducing external-impact risk. Points are deducted because it does not require user confirmation, disclose input data flow or sensitive financial-data handling, provide rollback, or address dependency security; Apache-2.0 and official publisher provenance provide limited attribution evidence but do not establish safety.
The document gives a reproducible basic Python/openpyxl path and a clear output contract, and distinguishes headless use from live Excel tooling. Points are deducted for missing dependency installation or pinning, input validation, exception handling, collision handling, formula recalculation checks, and diagnostic failure feedback; execution reliability cannot be verified statically, so the score is capped below 10.
The target scenario, environment distinction, and workbook output are reasonably clear, and the skill identifies when live Excel tools should be used instead. Points are deducted for lacking precise trigger semantics, non-fit boundaries, input-format constraints, Chinese-language workflow support, and mainland-China environment guidance; the core function does not entirely depend on overseas services.
The skill is well layered around an output contract, construction example, formatting conventions, and exclusion criteria; the repository supplies an Apache-2.0 license, validation CI, and contribution guidance. Points are deducted because the skill itself lacks versioning, changelog, named maintenance ownership or update path, and omits dependency setup, troubleshooting, limitation disclosure, and broader examples; repository CI does not demonstrate the skill's key paths.
The skill plausibly completes the core task of producing a basic Excel artifact headlessly and specifies a collectible path, input/formula separation, and modeling checks. Points are deducted because the example is simple and does not verify formula results, formatting integrity, complex models, or direct usability; outputs still require human review, and static review caps effectiveness at 7.
The source includes a concrete code example, output rules, repository revision, and related CI configuration, providing some auditability. Points are deducted because there are no committed xlsx-author tests, real execution results, key-path CI coverage, or independent third-party reproduction; the CI mainly validates plugin metadata rather than skill behavior, so the score remains below 5.
- Confirm that Python, openpyxl, and Bash are available, and pin or review dependency versions before use.
- The skill does not define storage, access-control, or cleanup handling for sensitive financial data; do not write restricted data to an unreviewed ./out/ directory.
- Writing formulas with openpyxl does not ensure that formulas have been calculated; recalculate and check results, circular references, and error values in a compatible spreadsheet engine before delivery.
- No recovery procedure is defined for existing files, script failure, or partial writes; the caller should implement safe temporary-file and rollback handling.
What it does & when to use it
xlsx-author is designed for headless execution, producing an Excel workbook as a file artifact instead of editing a live Office workbook. It saves the result to ./out/<name>.xlsx and uses openpyxl to create and edit workbooks. The skill fits managed-agent or CMA sessions where no Office application is open. It also defines conventions for inputs, formulas, links, named ranges, and model checks.
Writes and runs a short Python script using openpyxl to create or modify a workbook; creates sheets such as Inputs, calculation sheets, and Checks; marks hardcoded inputs, formulas, and links to other sheets or files in blue, black, and green; prevents hardcodes in calculation cells; creates named ranges for values referenced by a deck or memo; adds TRUE/FALSE checks for balance-sheet balances and cash-flow-to-cash ties; and saves one model as ./out/<name>.xlsx.
- A managed agent must deliver an Excel financial model from a server environment with no open Office application.
- A financial analyst needs inputs, formulas, and cross-sheet references written with consistent modeling colors.
- A model builder needs balance-sheet and cash-flow checks included before delivery.
- An automated workflow needs a .xlsx artifact that an orchestration layer can collect.
Pros & cons
- Explicitly supports managed-agent environments without an open Office application.
- Defines a clear output path and delivery contract.
- Includes financial-modeling conventions for colors, named ranges, and checks.
- Uses openpyxl, making it suitable for file-based automation workflows.
- Covers headless file generation rather than driving a live Excel workbook.
- Requires Python, Bash, and openpyxl; the source gives no version requirements.
- The source does not document a test suite, error-handling strategy, or broad template-compatibility guarantees.
- It requires one model per file and should not append to an existing workbook unless explicitly requested.
How to install
Add the repository as a Claude plugin marketplace: claude plugin marketplace add anthropics/financial-services. Install the core financial-analysis plugin: claude plugin install financial-analysis@claude-for-financial-services. The source skill is at plugins/vertical-plugins/financial-analysis/skills/xlsx-author/SKILL.md. The source does not document a separate installation command for this individual skill.
How to use
In a headless managed-agent or CMA session, make a request such as: “Use xlsx-author to generate an Excel financial model and save it to ./out/model.xlsx.” Run the required Python/openpyxl script, ensure the output path is ./out/<name>.xlsx, and return that relative path in the final message. The source does not define a dedicated trigger command or more specific prompt syntax.
Compared to similar skills
Compared with mcp__office__excel_* tools, xlsx-author targets headless environments and writes a workbook artifact to disk. When Office tools are available, the source recommends using them to drive a live workbook with review checkpoints.