What does this skill do, and when should you use it?
xlsx-author creates Excel workbook artifacts in headless managed-agent or CMA sessions. It uses Python and openpyxl, saving the result under ./out/<name>.xlsx. The skill specifies workbook conventions for inputs, formulas, links, named ranges, and balance checks. When live Office Excel tools are available, the source directs users to use those tools instead.
Creates ./out/ when needed; writes and runs a short Python script; builds, formats, and saves .xlsx workbooks with openpyxl; uses colors to distinguish hardcoded inputs, formulas, and links; keeps inputs on an Inputs sheet; adds named ranges for values referenced from decks or memos; and creates a Checks sheet that surfaces TRUE/FALSE checks.
- An investment banking or research professional needs an Excel file artifact from a managed-agent session with no Office app open.
- A financial modeler needs inputs, calculations, and checks separated across workbook tabs.
- An operations team needs to generate one Excel file programmatically with Python.
- An analyst needs named ranges for values referenced by a presentation or memo.
How do you install this skill?
- Before use, confirm that a compatible openpyxl version is installed and manually review formulas, formatting, checks, and sensitive-data exposure.
- Failure cleanup and rollback are unspecified; do not assume existing files may be overwritten or that financial data may be written to an uncontrolled directory.
- Repository-level CI does not demonstrate that this skill's workbook-generation path has been tested.
- Shell / CLI
- Local filesystem
Pythonopenpyxl
In Cowork, open Settings → Plugins → Add plugin, paste https://github.com/anthropics/financial-services, and select the required agents or verticals from the marketplace list. The README does not provide a standalone xlsx-author installation command; it is listed in the financial-analysis core plugin and appears as a bundled skill in Earnings Reviewer.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/anthropics/financial-services.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/plugins/agent-plugins/earnings-reviewer/skills/xlsx-author" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Use xlsx-author to create an Excel workbook and save it as ./out/model.xlsx.
In a headless managed-agent/CMA session, use a prompt such as “Use xlsx-author to create an Excel workbook and save it as ./out/model.xlsx.” The skill uses Python and openpyxl, then returns the relative artifact path in the final message. If mcp__office__excel_* tools are available, use the live-workbook workflow instead.
What are this skill's strengths and limitations?
- Defines a predictable ./out/<name>.xlsx output contract for orchestration and collection.
- Provides concrete conventions for inputs, formulas, links, named ranges, and checks.
- Designed for managed environments without an open Office application.
- Covers workbook production, not complete financial-model logic or data retrieval.
- Requires Python, openpyxl, shell access, and local filesystem access.
- Not intended for editing a live Excel workbook; the source provides no skill-specific test suite or platform validation record.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Headless Excel Workbook Builder this page ✓ Anthropic · Official | 51 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
| xlsx Spreadsheet Skill | 49 · Use with care | ★ 3.1k | 8mo ago | MIT |
| DeepTutor XLSX Skill | 60 · Recommended | ★ 41k | 17d ago | Apache-2.0 |
| XLSX Spreadsheet Skill ✓ Anthropic · Official | 40 · Not recommended | ★ 181k | 16d ago | — |
| OfficeCLI Data Dashboard Skill | 57 · Use with care | ★ 32k | 12d ago | Apache-2.0 |
Compared with the mcp__office__excel_* live Excel workflow, xlsx-author targets headless sessions and produces a file artifact. The source provides no other named alternatives for comparison.
How did FollowSkills review this skill?
The skill limits itself to headless execution and specifies writing to ./out/<name>.xlsx; it does not request network access, credentials, or external side effects, and it prohibits overwriting an existing workbook unless explicitly asked. It lacks guidance for sensitive financial data, user confirmation, failure rollback, and dependency security, so points are deducted.
The instructions and example are internally consistent, using openpyxl with explicit input, formula, and check-sheet conventions. However, dependencies are not pinned, there are no key-path tests or execution artifacts, and abnormal inputs, save failures, and formula errors lack diagnostic feedback. The static ceiling and these gaps require deductions.
The target scenario, output form, and non-fit for live Excel/Cowork mode are reasonably clear. Trigger semantics, input boundaries, applicability beyond financial models, and Chinese-environment support are not specified, so only a mid-range score is justified.
The document is readable and layered, with an output contract, build example, formatting conventions, and a non-use section. The repository supplies an Apache-2.0 license, CI validation, and official publisher provenance, but the skill lacks versioning, changelog, explicit maintenance ownership, dependency installation notes, and troubleshooting guidance.
The skill can directly guide creation of a collectable .xlsx artifact and provides a reusable openpyxl example plus modeling conventions. It does not verify representative outputs, establish complete business-result correctness, or compare marginal benefit with manual or alternative workflows; human review remains necessary, so the static score is capped below 7.
The source includes a clear output contract, code example, repository-level plugin validation, and secret scanning, providing limited auditable evidence. It lacks skill-specific key-path tests, committed output samples, or third-party execution evidence, so a higher score is not supported.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →