Headless Excel Workbook Builder
Generate deliverable Excel files when no Office application is available.
The skill scopes output to ./out/<name>.xlsx and specifies openpyxl, with no evident malicious external effects; however, it lacks user confirmation, sensitive-data handling, dependency-security guidance, rollback, and explicit data-flow disclosure, so points are deducted.
The headless context, output path, openpyxl example, and modeling conventions make the happy path plausible; however, dependencies are neither pinned nor installed, and there is no abnormal-input handling, failure feedback, overwrite protection, or committed testing, warranting a conservative score.
The intended managed-agent/CMA scenario is clear, and the non-fit condition for available live Excel tools is stated; inputs, naming parameters, trigger boundaries, complex-workbook limits, and Chinese-language support are insufficiently specified, so the score is limited.
The document has clear purpose, output contract, example, conventions, and a non-use section; the repository supplies an Apache-2.0 license and verified official provenance, but the skill lacks versioning, changelog, named maintenance responsibility, update path, troubleshooting, and FAQs.
It can direct creation of a collectible .xlsx artifact and provides useful input/formula/check conventions; representative outputs, formula correctness, formatting integrity, and compatibility are not verified, and human review remains necessary, so the static score is conservative.
Repository CI validates plugins and scans for secrets, but it does not cover the skill's workbook-generation key paths; the skill contains only prose and a code example, with no committed tests, execution evidence, or independent corroboration.
- Confirm that a compatible Python environment and openpyxl are installed; installation and version requirements are unspecified.
- Same-named outputs may be overwritten and formulas may be incorrect; manually review the workbook, formulas, checks, and sensitive data before delivery.
- The skill is intended for headless use when live Excel tools are unavailable; Chinese-language support, localized templates, and complex Excel feature compatibility are unproven.
What it does & when to use it
xlsx-author is designed for headless managed-agent or CMA sessions that need to produce an .xlsx file instead of editing an open Excel workbook. It requires the workbook to be saved under ./out/<name>.xlsx and the relative path to be returned in the final response. The skill uses openpyxl and defines conventions for inputs, formulas, links, named ranges, and balance checks. It fits automated file-producing workflows, but not workflows centered on live Excel interaction.
It instructs the agent to write and run a short Python script with openpyxl, create workbook sheets and cells, add formulas, apply font colors, and save the file to the required output directory. Hardcoded inputs are blue, formulas are black, and links to another sheet or file are green; calculation cells must not contain hardcodes, values referenced from a deck or memo should use named ranges, and a Checks sheet should expose TRUE/FALSE balance checks. The resulting file is written to ./out/<name>.xlsx and the relative path is returned.
- A managed-agent user needs to deliver an Excel financial model in an environment without an open Office application.
- A financial analyst needs to generate a workbook with Inputs, calculation sheets, and a Checks sheet using Python and openpyxl.
- An automation pipeline needs the generated .xlsx file as an artifact that a downstream orchestration layer can collect.
- A model builder needs consistent conventions for hardcoded inputs, formulas, cross-sheet links, and balance checks.
Pros & cons
- Explicitly supports managed-agent environments without an Office application.
- Produces a collectible .xlsx file artifact with a defined output path.
- Provides concrete conventions for inputs, formulas, cross-sheet links, named ranges, and checks.
- Uses Python and openpyxl, making scripted workbook generation practical.
- Requires an environment that can run Bash and Python, plus the openpyxl dependency.
- Does not drive a live Excel workbook; the source directs users to mcp__office__excel_* when those tools are available.
- The source provides no test suite, version requirements, or failure-recovery procedure.
- The skill defines workbook-generation conventions but does not document support for complex templates, charts, or macros.
How to install
In Cowork, open Settings → Plugins → Add plugin, paste https://github.com/anthropics/financial-services, and select Pitch Agent; or use Claude Code: run claude plugin marketplace add anthropics/financial-services, followed by claude plugin install pitch-agent@claude-for-financial-services. The source does not document a separate xlsx-author installation command.
How to use
In a headless managed-agent or CMA session, request an Excel file, for example: “Create a financial model with Inputs, calculation, and Checks sheets, and save it as ./out/model.xlsx.” The agent should use Python and openpyxl, save the output as ./out/<name>.xlsx, and return the relative path in its final response. No dedicated trigger phrase is documented.
Compared to similar skills
Compared with using mcp__office__excel_* to operate a live Excel workbook, xlsx-author targets headless environments and directly produces a file artifact; the source recommends the live-workbook approach when Office tools are available.