hashgraph-online/awesome-codex-plugins / plugins/alpoxdev/hypercore/skills/hyper-excel
hyper-excel
VerifiedUse this skill when the user asks to read, analyze, create, or modify Excel workbooks (.xlsx, .xlsm) or turn CSV data into spreadsheets - inspecting sheets and formulas, editing cells while keeping existing formatting and formulas intact, building formatted reports with charts, or auditing formulas for broken ranges and error values. Do not use for plain dataframes, PDF or slide output, or VBA macro authoring.
references/research-2026-11-13.md
View on GitHub# hyper-excel research record
> Checked 2026-11-13. Every number was measured the same day with `gh api`/`curl`.
## Top 20 candidates
| # | Repository · path | Stars | Last push | License | What this skill took |
|---|---|---|---|---|---|
| 1 | [anthropics/skills](https://github.com/anthropics/skills) — `xlsx` | 193,971 | 2026-11-06 | none (no license file in the repository) | makes formula recalculation a mandatory pipeline step |
| 2 | [anthropics/claude-code](https://github.com/anthropics/claude-code) | 155,524 | 2026-11-13 | none | skill trigger wording and how scope is described |
| 3 | [anthropics/claude-cookbooks](https://github.com/anthropics/claude-cookbooks) | 54,601 | 2026-11-04 | MIT | layout of data in/out examples |
| 4 | [pandas-dev/pandas](https://github.com/pandas-dev/pandas) | 50,506 | 2026-11-12 | BSD-3-Clause | what `to_excel` loses (openpyxl engine) |
| 5 | [wshobson/agents](https://github.com/wshobson/agents) | 42,471 | 2026-11-09 | MIT | description length and separation from triggers |
| 6 | [anthropics/financial-services](https://github.com/anthropics/financial-services) — `xlsx-author` | 41,267 | 2026-11-12 | Apache-2.0 | assumption-sheet separation for financial models |
| 7 | [SheetJS/sheetjs](https://github.com/SheetJS/sheetjs) | 36,345 | 2024-04-18 | Apache-2.0 | part-oriented read model |
| 8 | [davila7/claude-code-templates](https://github.com/davila7/claude-code-templates) | 33,464 | 2026-11-13 | MIT | skill distribution layout |
| 9 | [iOfficeAI/OfficeCLI](https://github.com/iOfficeAI/OfficeCLI) | 33,259 | 2026-11-12 | Apache-2.0 | CLI subcommand surface design |
| 10 | [VoltAgent/awesome-claude-code-subagents](https://github.com/VoltAgent/awesome-claude-code-subagents) | 26,527 | 2026-11-02 | MIT | subagent role separation |
| 11 | [qax-os/excelize](https://github.com/qax-os/excelize) | 21,136 | 2026-11-13 | BSD-3-Clause | streaming row writes |
| 12 | [genspark-ai/genoffice](https://github.com/genspark-ai/genoffice) | 12,292 | 2026-11-13 | Apache-2.0 | one entry point across office documents |
| 13 | [haris-musa/excel-mcp-server](https://github.com/haris-musa/excel-mcp-server) | 4,297 | 2026-11-10 | MIT | tool naming and argument design (`create_workbook`, `apply_formula`) |
| 14 | [jmcnamara/XlsxWriter](https://github.com/jmcnamara/XlsxWriter) | 3,994 | 2026-10-28 | BSD-2-Clause | limits of a write-only design and chart serialization |
| 15 | [spiculedata/saiku](https://github.com/spiculedata/saiku) | 1,339 | 2026-11-13 | Apache-2.0 | reducing pivot and OLAP aggregation to ordinary rows |
| 16 | [negokaz/excel-mcp-server](https://github.com/negokaz/excel-mcp-server) | 1,049 | 2025-07-19 | MIT | splitting tools by sheet and range |
| 17 | [sbroenne/mcp-server-excel](https://github.com/sbroenne/mcp-server-excel) | 887 | 2026-11-13 | MIT | cell and conditional-format manipulation surface |
| 18 | [tfriedel/claude-office-skills](https://github.com/tfriedel/claude-office-skills) | 849 | 2026-04-01 | none | separating Excel automation skill from xlsx manipulation |
| 19 | [claude-office-skills/skills](https://github.com/claude-office-skills/skills) | 574 | 2026-01-31 | MIT | office skill catalog structure |
| 20 | [alchaincyf/huashu-excel](https://github.com/alchaincyf/huashu-excel) | 464 | 2026-11-04 | MIT | the "inspect → clean → analyze → deliver" staging |
Smaller repositories consulted as well:
| Repository | Stars | Last push | License | What was worth noting |
|---|---|---|---|---|
| [BayramAnnakov/excel-hygiene](https://github.com/BayramAnnakov/excel-hygiene) | 18 | 2026-06-03 | Apache-2.0 | turning the Panko–Halverson error taxonomy into deterministic checks |
| [petehottelet/spreadsheet-auditor](https://github.com/petehottelet/spreadsheet-auditor) | 11 | 2026-10-05 | MIT | how audit report items are organized |
| [officecli/officecli](https://github.com/officecli/officecli) | 114 | 2026-08-10 | MIT | single-binary office tool surface |
## Analysis of the top 5
### 1. anthropics/skills `xlsx`
Its `SKILL.md` forces "read → modify → recalculate → verify" in that order, and `scripts/recalc.py` splits writing formula results into the cache out as a separate script. **Taken**: recalc as a required pipeline step, not an option. **Not taken**: requiring LibreOffice as the recalculation engine — this machine has none, and not requiring an install was part of the brief, so an own evaluator replaced it.
### 2. anthropics/financial-services `xlsx-author`
It gathers assumptions into one sheet and makes every calculation reference those cells. **Taken**: the design instruction to keep assumptions in one sheet and reference them from calculations (reflected in this skill's design-first workflow and in the `hyper-excel-workflow-design-first-ko` case).
### 3. haris-musa/excel-mcp-server
It exposes tools such as `create_workbook`, `apply_formula`, `validate_formula_syntax`, `create_chart`, and `create_pivot_table` from one server. The README advertises "PivotTables, slicers". **Taken**: splitting the tool surface by user intent. **Not taken**: native pivot generation — this skill works on OOXML directly, and hand-writing a pivot cache is a risk it declines; it writes computed rows instead and says so.
### 4. iOfficeAI/OfficeCLI
One binary handles Word, Excel, and PowerPoint, with subcommands split into `read/edit/automate`. **Taken**: naming subcommands after user verbs (read, verify, create, edit). This skill's `inspect/verify/build/patch/recalc/profile` is the result.
### 5. BayramAnnakov/excel-hygiene
It implements the 13-item Panko–Halverson spreadsheet error taxonomy as deterministic checks in `check_excel.py`. **Taken**: pinning error types as codes and reporting by code. The 12 items in `rules/failure-modes.md` and the 8 finding codes in `verify_formulas.py` started here.
## Licensing
Every repository contributed ideas only; no code was copied. anthropics/skills and anthropics/claude-code carry no license file, so copying from them was not possible in the first place. The rest are MIT/Apache-2.0/BSD, but this skill's implementation is an independent one that handles OOXML with the standard library.
## How this skill differs
| Axis | Typical Excel skill | hyper-excel |
|---|---|---|
| Dependencies | openpyxl/pandas required | standard library only; openpyxl is an optional read-back check |
| Editing | re-save the whole file | part-level patch with a SHA-256 proof for untouched parts |
| Recalculation | requires LibreOffice/Excel | own evaluator; unsupported functions are reported as unevaluated, never guessed |
| Pivot | attempts a native pivot | computed rows plus an autofilter range, and states that it is not native |
| Korean | mentioned generally | cp949, BOM, `2026. 1. 31.`, and `#,##0"원"` backed by measurements |
## Sources
> Checked 2026-11-13.
| Claim | Source |
|---|---|
| Star counts, last push dates, and licenses for every repository in the tables | `gh api repos/<owner>/<repo> --jq '[.stargazers_count, .pushed_at, .license.spdx_id]'`, run 2026-11-13 |
| Candidate discovery | `gh api 'search/repositories?q=xlsx+skill+in:name,description&sort=stars'` and `...?q=excel+mcp&sort=stars`, run 2026-11-13 |
| openpyxl version 3.1.5 and its source host | `https://pypi.org/pypi/openpyxl/json`, read 2026-11-13 |
| `xlsx` pipeline and `scripts/recalc.py` | <https://github.com/anthropics/skills> (path `xlsx`) |
| `xlsx-author` assumption-sheet structure | <https://github.com/anthropics/financial-services> (the `xlsx-author` folder in its financial-analysis plugin) |
| `excel-hygiene` error taxonomy checks | <https://github.com/BayramAnnakov/excel-hygiene> |
### Evidence grade
`LOCAL` — every number was measured on 2026-11-13 with the command in the table. Star counts change daily; the date is part of the claim. Repository contents were read, not executed, so the "what this skill took" column is a design judgement, not a measurement.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |