Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/zhanglunet/openworker-zh-localizednpx agentmods add skills/zhanglunet/openworker-zh-localized/excel-ai-analystWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/zhanglunet/openworker-zh-localized/excel-ai-analyst)<a href="https://agentmods.dev/skills/zhanglunet/openworker-zh-localized/excel-ai-analyst"><img src="https://agentmods.dev/badge/skills/zhanglunet/openworker-zh-localized/excel-ai-analyst/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/zhanglunet/openworker-zh-localized/excel-ai-analyst"><img src="https://agentmods.dev/badge/skills/zhanglunet/openworker-zh-localized/excel-ai-analyst.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00219 | $0.05248 |
| Opus 5 | $0.00110 | $0.02624 |
| Sonnet 5 | $0.00044 | $0.01050 |
| Haiku 4.5 | $0.00022 | $0.00525 |
Grade A, and why
excel-ai-analyst scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 280 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Excel 智能分析（把表格当代码读）
核心主张
一张成熟的业务 Excel 就是一套没有文档的遗留代码：列名是变量名，公式是函数， 跨 Sheet 引用是模块依赖，check 列是单元测试。所以分析它的正确姿势不是"看数据"， 而是逆向工程：先还原变量定义（本体）→ 再还原函数与调用图（公式链 + 血缘）→ 然后用全量真实数据跑一遍回归测试（验证）→ 通过了才有资格谈分析和推演。
跳过验证直接分析 = 在没读懂代码的情况下改代码。 这是本技能最重要的一条纪律。 用户催"直接给结论"时也不能跳——没验证过的结论，错了没人知道。
五步法
| 步 | 产出 | 谁来做 |
|---|---|---|
| 1 结构化转换 | 每个 Sheet 一份 MD：多行表头、合并单元格、真实公式、列画像 | excel_ai.py tomd |
| 2 字段本体 | 跨表统一的字段定义表（含义/类型/计算关系/来源） | AI 手写 |
| 3 公式链与血缘 | 分层公式链 + 跨 Sheet 数据流向 | AI 手写 |
| 4 数据验证 | 逐行比对「AI 理解」vs「Excel 实际值」，出通过率与不匹配清单 | excel_ai.py verify |
| 5 交付 | 带配色标注的结果表 + 分析报告（分布/异常/What-If） | excel_ai.py output / analyze |
Step 2/3 是认知工作，必须由你读着 Step 1 的 MD 亲自写，不要试图用脚本自动生成—— 业务含义不在文件里，在列名的言外之意里。Step 1/4/5 是机械工作，交给脚本。
开工第一件事是 todo_write 把这五步列出来（用户的进度面板就是它渲染的），
每完成一步更新状态，同时只保持一项 in_progress。
开工前：三件事
1. 找到脚本
脚本随技能分发，就在 load_skill 返回的 resources_path 下：
<resources_path>/scripts/excel_ai.py
用 load_skill 返回的那个 resources_path，不要假设当前目录下有 scripts/。
（技能装在全局技能目录或项目技能目录里，与用户的工作区不是同一个地方；
一般是 ~/.config/coworker/skills/excel-ai-analyst，Windows 在 %APPDATA%\coworker\skills\ 下，
项目级则在 <工作区>/.coworker/skills/ 下——路径会变，resources_path 不会错。）
先确认它在：run_shell 跑 ls <resources_path>/scripts，或 python3 <resources_path>/scripts/excel_ai.py --help。
万一确实缺文件（技能被裁剪过），按 references/spec-schema.md 的「附录」现写一份，别硬着头皮手算。
2. 让 python3 能跑
OpenWorker 的命令执行走审批：run_shell 默认每次弹审批卡，除非命令命中用户全局配置里的
allowed_commands 白名单。第一次调用时告诉用户：
想免掉后面每一步的审批，可以在全局配置
~/.config/coworker/config.toml（Windows：%APPDATA%\coworker\config.toml）里加一行：allowed_commands = ["python3"]更保守的企业写法是只放行这一个脚本（白名单按命令词前缀匹配）：
allowed_commands = ["python3 /Users/你/.config/coworker/skills/excel-ai-analyst/scripts/excel_ai.py"]白名单只在用户全局配置里生效；工作区
.coworker/config.toml里的同名配置要等这个工作区被信任后才追加。
两条会让白名单失效、必然弹审批的写法，能避就避：
- 命令里带
&&、;、|、>、<、`、$(、(或换行——一律强制审批。所以 一条命令只做一件事，不要串联、不要重定向。 - 文件名带括号（
工资表(1).xlsx）也会命中上面这条。必要时先把表复制成不带括号的名字。
What ships with it
17 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- references/pitfalls.md 9.0 KB
- references/spec-schema.md 17 KB
- references/walkthrough.md 14 KB
- scripts/excel_ai.py 163 KB runs code
- tests/attacks/dirty-data/make_attack_fixtures.py 14 KB runs code
- tests/attacks/dirty-data/run_attacks.py 33 KB runs code
- tests/attacks/security/_runner.py 2.0 KB runs code
- tests/attacks/security/a1_probe_eval.py 6.2 KB runs code
- tests/attacks/security/a2_spec_cli.py 13 KB runs code
- tests/attacks/security/a3_csv_injection.py 10 KB runs code
- tests/attacks/security/a4_perf.py 8.2 KB runs code
- tests/attacks/security/run_all.py 1.6 KB runs code
- tests/attacks/spec-robustness/make_fixtures.py 4.9 KB runs code
- tests/attacks/spec-robustness/run_attacks.py 35 KB runs code
- tests/make_fixtures.py 25 KB runs code
- tests/README.md 2.8 KB
- tests/test_excel_ai.py 55 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 280 lines · 219 tokens per session scan A ebb97c68e606
excel-ai-analyst is a skill published in the GitHub repository zhanglunet/openworker-zh-localized (2 stars, last pushed 4d ago), licensed MIT. It adds 219 tokens to every session and 5,248 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
chat-complex-documents
Chat with and search your complex documents — ask questions, extract tables and fields, and get answers grounded in the source. Connects the hosted Unstructured Transform MCP server to parse, structure, and enrich PDFs, Word/Excel/PowerPoint, images, scanned files, emails, and 60+ other formats into clean, AI-ready…
google-apps-script
Build Google Apps Script automation for Sheets and Workspace. Custom menus, triggers (onEdit / time-driven / form submit), dialogs, sidebars, email batches, PDF export, external API. Use whenever the user wants to automate a Google Sheet, build a Sheets menu / sidebar / dialog, hit a Sheets row from email or a…
office-document-editor
Edit existing Office documents in the project space. Use when the user asks to (1) modify a PPT slide's title/content/data, (2) update a Word document's section/paragraph/table, (3) change Excel cell values/formulas/rows/columns, (4) add/remove/reorder slides or pages, (5) fix formatting, update data, or make…
Google Workspace integration — Gmail, Drive, Calendar, Contacts, Tasks, and Sheets via gogcli.
spreadsheet
Read, create, modify, analyze, and convert spreadsheet files (xlsx, csv, tsv).
library
A skill for managing WorkBuddy’s built-in library, which stores online documents, web links, data tables, dashboards, web pages, files, and shared spaces. It also supports publishing, collaboration, and file transfers.