[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7DBPCT861S4PZDXJ0E712S6":3,"skill-same:01M7DBPCT861S4PZDXJ0E712S6":302},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":31,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M7DBPCT861S4PZDXJ0E712S6","xlsx",null,"Spreadsheet skill — read, edit, create, and convert .xlsx\u002F.xlsm\u002F.csv\u002F.tsv files. Trigger when a spreadsheet file is the primary input or output: editing columns, formulas, formatting, charting, cleaning messy data, or creating new spreadsheets. Not for Word\u002FHTML\u002FPDF deliverables even if tabular data is involved.","ZCodium-project\u002FZCodium","ZCodium-project","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F331969751?v=4","01M7DBKVDM2CY9G27NC8F39MCG","ZCodium","01M7DBKVEPNTJN0R8FNRV88JEH",253,"MIT","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: xlsx\ndescription: >-\n  Spreadsheet skill — read, edit, create, and convert .xlsx\u002F.xlsm\u002F.csv\u002F.tsv files.\n  Trigger when a spreadsheet file is the primary input or output: editing columns, formulas, formatting, charting, cleaning messy data, or creating new spreadsheets.\n  Not for Word\u002FHTML\u002FPDF deliverables even if tabular data is involved.\ndescriptions:\n  zh-Hans: \"读取、编辑、创建和转换表格文件，支持 xlsx、xlsm、csv、tsv、公式、格式、图表和数据清洗。\"\nlicense: MIT\n---\n\n\u003C!--\nCopied from MiniMaxAI\u002Fminimax-code (packages\u002Flocal-runtime\u002Fassets\u002Fskills\u002Fxlsx)\nat revision 564e9166d81f87b0b767b005e4779d4697b512be.\nCopyright (c) 2026 MiniMax Code. Licensed under MIT.\nModified by ZCode: added the provenance notice; replaced a pointer into a different skill in the upstream host product with a self-contained Windows tool table (detect + install command), and stated that this fork ships no auto-installer. All other skill instructions unchanged.\nSee THIRD-PARTY-NOTICES.md in the repository root for license and provenance.\n-->\n\n# xlsx\n\nA pragmatic, recipe-first guide for reading, editing, creating, and\nrecalculating `.xlsx` \u002F `.xlsm` \u002F `.csv` \u002F `.tsv` files. The default\npairing is **pandas for tabular data + openpyxl for formulas, styles,\nnamed ranges, and charts**. Recalculation through\n[`scripts\u002Frecalc.py`](scripts\u002Frecalc.py) (LibreOffice headless) is\nmandatory before delivery — openpyxl writes formulas as strings and\nnever evaluates them.\n\n> **Formula-first.** A spreadsheet without live formulas is just a\n> CSV with a fancier extension. **Every computed value** — totals,\n> averages, growth rates, ratios, cross-sheet references, percent\n> changes, anything derivable from other cells — **must be written as\n> a live `=…` formula**, not as the pre-computed number. Hard-coded\n> numbers belong only in the **Assumptions** block (inputs the user\n> can flip to re-run the model). When in doubt, write the formula.\n> Full convention in §5 and\n> [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §4.\n>\n> **❌ The most common anti-pattern.** Loading a workbook with pandas,\n> computing derived columns in Python (`df[\"total\"] = df[\"a\"] + df[\"b\"]`,\n> `df.groupby(...).sum()`, etc.), then writing the result back with\n> `df.to_excel(...)` ships **static numbers** — the workbook becomes a\n> dead snapshot the moment any input changes. Use pandas\u002Fpolars only\n> to load and clean **raw inputs**; emit every derived value as a live\n> `=` formula via openpyxl (§3.2). This rule applies regardless of how\n> easy it would be to compute the value in Python first.\n\n> **Spreadsheet output only.** For Word documents, PowerPoint slides,\n> HTML reports, standalone Python scripts, database pipelines, or the\n> Google Sheets API, switch to the matching skill.\n\n## Operational rules — read before doing anything\n\n> **1. Match user query against [`docs\u002Fpitfalls-index.md`](docs\u002Fpitfalls-index.md)\n> FIRST.** It contains 8 production-ready **canonical query templates**\n> (X1–X8), each with a `Match signatures` block (sample queries) and a\n> complete executable prompt that already encodes Formula-first, recalc\n> verification, formula-count gate, sample-data-integrity, slicer\n> handling, and every other rule below. Workflow:\n>\n> 1. Scan the Quick lookup table — match user's query keywords to a row.\n> 2. **Copy the matching canonical query verbatim**, substitute the\n>    `Slots` (e.g. `{INPUT}`, `{COMPANY}`, `{OUTPUT_XLSX}`) with the\n>    user's actual values, and execute step-by-step.\n> 3. Multiple partial matches → fuse: take the strictest verification\n>    from each, never relax a constraint.\n> 4. No match → fall back to the Decision Tree \u002F per-section guides below.\n>\n> Do NOT skip verification steps in the canonical queries — every\n> \"ship a wrong workbook\" failure traces back to skipping recalc,\n> total_formulas check, or row-count canary.\n\n> **2. Formula-first is non-negotiable.** A delivery that ships static\n> numbers where formulas were possible is a **failed delivery** — even\n> if every number is numerically correct the moment you saved it. The\n> user's first edit will expose the lie. Hardcode only the Assumptions\n> block; everything derivable from other cells goes in as `=…`. Full\n> regime in §5.\n\n> **3. `total_formulas == 0` is a red flag, not a green light.** A\n> \"success\" recalc with zero formulas means the workbook is a static\n> dump — the user can't audit, can't re-run scenarios, can't\n> introspect derivations. Treat it as a delivery failure and rewrite\n> via §3.2 \u002F openpyxl `=` formulas. The only legitimate exception is\n> a workbook the user explicitly asked to be a static snapshot\n> (in which case `pandas.to_excel` is the right tool, not this skill).\n\n> **4. Source-data-integrity rule — never `df.sample(N)` \u002F `df.head(N)`\n> on the raw sheet.** Down-sampling 400k rows to 100k \"because openpyxl\n> writes are slow\" silently destroys every aggregation built on top —\n> the pivot or `=SUMIF` summary will be off by ~75% and look entirely\n> plausible. **Write the full row count, even if it takes minutes.**\n> If write throughput is the real bottleneck, switch the writer\n> (xlsxwriter §3.3, or `Workbook(write_only=True)` streaming pattern\n> in [`docs\u002Fadvanced-reference.md`](docs\u002Fadvanced-reference.md) §6),\n> never the sample size.\n\n> **5. Summaries must be Excel-native — pivot tables or `=SUMIFS` \u002F\n> `=COUNTIFS` over the Raw sheet, never Python `groupby` written back\n> as values.** This is the same rule as §3 phrased for the most\n> common offender. The user expects: edit a Raw row, hit recalc, see\n> the totals move. Static summaries break that contract silently.\n\n> **6. Slicers cannot be authored from scratch by openpyxl.** The\n> `xl\u002Fslicers\u002F*.xml` part requires GUID-bound pivot cache references\n> that openpyxl has no API for. Two production paths:\n> (a) template-inheritance — author a `template.xlsx` once in Excel\n> or LibreOffice with the slicer wired to a named range, then\n> `load_workbook(template) → write into named range → save as new`;\n> (b) raw-XML transplant from a known-good template via\n> `scripts\u002Foffice\u002F{unpack,pack}.py`. **If the user asks for a slicer\n> and you have no template, say so before silently downgrading to a\n> static dropdown.**\n\n> **7. Don't suppress stderr.** `2>\u002Fdev\u002Fnull` is **never** the right\n> choice in this skill (`recalc.py`, `soffice`, `pdfinfo`, `unpack.py`,\n> `pack.py`). On failure you lose the only signal that explains why\n> and have to rerun blind. If output is too noisy, redirect to a log\n> file and grep on demand:\n>\n> ```bash\n> python scripts\u002Frecalc.py file.xlsx 60 2>\u002Ftmp\u002Frecalc.log\n> # If JSON shows status != \"success\" or exit != 0, then:\n> #   grep -in \"error\\|trace\\|fail\" \u002Ftmp\u002Frecalc.log | head -20\n> ```\n\n> **8. When the user states a numeric range for a derived value,\n> enforce it in the formula.** \"Composite score 0–100\" → wrap the\n> formula with `=ROUND(MIN(MAX(raw, 0), 100), 1)`. Don't ship a\n> 0–1804 column and blame \"data anomaly\". Spot-check `MIN()` \u002F\n> `MAX()` of the output column after recalc.\n\n> **9. 500k+ rows: pandas\u002Fpolars first, not openpyxl row walking.**\n> For large tabular inputs (roughly **500,000+ rows**, or hundreds of MB),\n> default to `pandas`\u002F`polars` for reading, filtering, type normalization,\n> joins, and QA spot-checks. Do **not** iterate cell-by-cell with openpyxl to\n> inspect or transform the source workbook — it is too slow and encourages\n> accidental sampling. Use openpyxl only at the output boundary for formulas,\n> styles, charts, templates, and final `.xlsx` assembly. If the output is a\n> static raw-data dump, `DataFrame.to_excel`\u002F`xlsxwriter` is acceptable; if it\n> contains derived values, write full raw rows and Excel-native `=` formulas.\n\n> **10. Non-standard XLSX packages: trust recalc raw-XML fallback.**\n> Workbooks with heavy merged cells, charts\u002Fdrawings, or vendor-generated XML\n> can make openpyxl crash even after LibreOffice recalculated successfully.\n> `scripts\u002Frecalc.py` now catches openpyxl parse failures and falls back to a\n> raw worksheet XML scanner. If JSON returns `scanner: \"raw_xml_fallback\"` with\n> `status: success` \u002F `errors_found`, treat it as authoritative and do not\n> retry blindly. Use the `compatibility_hint` field to decide whether to avoid\n> downstream openpyxl rewrites. Details in [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §10.\n\nFor deeper material — creation and edit recipes, recalc internals,\nfinancial-model conventions, large-file alternatives — see\n[`docs\u002F`](docs\u002F).\n\n---\n\n## 1. Scope and When to Use\n\nThis skill covers the everyday spreadsheet chores an LLM agent is most\noften asked to perform end-to-end:\n\n- inspecting an existing workbook (sheet names, merged regions, headers)\n- reading tabular data into pandas \u002F polars for analysis or QA\n- creating a new workbook with formulas, styles, named ranges, and charts\n- editing an existing workbook in place (insert \u002F delete \u002F restyle \u002F replace)\n- recalculating every formula via [`scripts\u002Frecalc.py`](scripts\u002Frecalc.py)\n  and verifying `total_errors == 0` in the JSON\n- enforcing the financial-model conventions in §5 on every delivery\n- converting between tabular formats (`csv` ↔ `xlsx` ↔ `ods` ↔ `tsv`)\n\nUse a different approach when:\n\n- the deliverable is a Word document, slide deck, standalone script, or\n  database pipeline — switch to the matching skill (`docx`, `pptx`, etc.)\n- the workflow is read-only and the user just wants the cell values\n  printed — `extract-text` (`docs\u002Fadvanced-reference.md` §5) is enough\n- the workbook lives behind the Google Sheets API rather than on disk\n\n---\n\n## 2. Decision Tree\n\n```\nWhat does the user want from \u002F for this workbook?\n|\n+-- Read tabular data only ───────────────> pandas \u002F polars     -> §3.1, §3.4\n+-- Inspect quickly without code ─────────> extract-text CLI    -> docs\u002Fadvanced-reference.md §5\n+-- Create a new workbook ────────────────> openpyxl            -> §3.2\n+-- Edit an existing workbook in place ───> openpyxl            -> §3.2\n+-- Output has any derived\u002Fcomputed cells > openpyxl + `=` formulas -> §3.2, §5\n+-- Very large workbook (500k+ rows) ─────> pandas\u002Fpolars read + openpyxl\u002Fxlsxwriter output -> §3.1, §3.4, docs\u002Fadvanced-reference.md §6\n+-- Recalculate formulas (mandatory) ─────> scripts\u002Frecalc.py   -> §4.1\n+-- Convert format (csv ↔ xlsx ↔ ods) ────> pyexcel             -> §3.5\n+-- Sandboxed env (no AF_UNIX) ───────────> office.soffice shim -> §4.1\n```\n\n**Default rule.** Read with pandas, write and edit with openpyxl,\nrecalculate with `scripts\u002Frecalc.py`. Drop to xlsxwriter when the\nworkload is write-only and throughput-bound; drop to polars when the\ninput file is too big for pandas to hold in memory. **Never use\n`pandas.to_excel` \u002F `polars.write_excel` for derived\u002Fcomputed values\n— those write static numbers, not live formulas. See §3.1's ⚠️ box.**\n\n---\n\n## 3. Library Cookbook\n\nFive subsections, one per library. Each entry is a minimum viable\nexample plus a short note. Deeper recipes — full styling, conditional\nformatting, charts, edit gotchas — live in\n[`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md).\n\n### 3.1 pandas — tabular read \u002F write (default)\n\n```python\nimport pandas as pd\n\nframe = pd.read_excel(\"mau_forecast.xlsx\", engine=\"openpyxl\")           # first sheet\nsheets = pd.read_excel(\"mau_forecast.xlsx\", sheet_name=None)            # {sheet: DataFrame}\nframe.to_excel(\"clean_mau.xlsx\", sheet_name=\"clean\", index=False)\n```\n\n> ⚠️ **`to_excel()` writes static numbers only — no formulas survive.**\n> Use it strictly for **raw data dumps**: cleaned input data, exported\n> query results, anything where every cell is itself a primary value.\n> The moment the deliverable contains a derived value — totals,\n> averages, ratios, growth rates, anything computable from other cells\n> — switch to §3.2 and write it as a live `=` formula. **Never** do\n> `df[\"total\"] = df[\"a\"] + df[\"b\"]; df.to_excel(...)`: the workbook\n> will look fine until the user edits an input and discovers the\n> \"totals\" don't move. For aggregations across rows\u002Fsheets, write the\n> raw rows here and emit `=SUM(...)` \u002F `=SUMIF(...)` on the openpyxl\n> side (`docs\u002Fadvanced-reference.md` §6 has the worked large-file\n> example).\n\nDefault for any \"read tabular data\" job. For 500k+ rows, prefer pandas\u002Fpolars\nfor source inspection, filtering, joins, and QA; avoid openpyxl cell-by-cell\nsource traversal. Pair with openpyxl only on the write side when you need\nformulas or formatting.\n\n### 3.2 openpyxl — create and edit (default for write + edit)\n\n```python\nfrom openpyxl import Workbook, load_workbook\nfrom openpyxl.styles import Font\n\nbook  = Workbook()                                                       # create\nsheet = book.active\nsheet[\"A1\"], sheet[\"B1\"], sheet[\"C1\"], sheet[\"D1\"], sheet[\"E1\"] = (\n    \"Quarter\", \"MAU (mm)\", \"Tokens\u002FMAU\", \"Take-rate\", \"ARR (¥mm)\"\n)\n# Inputs (Assumptions) — hardcoded primary values; §5.2 requires blue.\ninput_font = Font(color=\"0000FF\")\nsheet[\"A2\"], sheet[\"B2\"], sheet[\"C2\"], sheet[\"D2\"] = \"2025Q3\", 148, 27, 0.85\nfor ref in (\"B2\", \"C2\", \"D2\"):\n    sheet[ref].font = input_font\n# Derived ARR — references inputs, not literal numbers; §5.2 requires black.\nsheet[\"E2\"] = \"=B2*C2*D2\"\nsheet[\"E2\"].font = Font(color=\"000000\")\nbook.save(\"mau_forecast.xlsx\")\n\nbook = load_workbook(\"mau_forecast.xlsx\")                               # edit\nbook[\"Sheet\"][\"D2\"] = 0.90                                              # flip take-rate\n# E2 still says \"=B2*C2*D2\" — recalc.py will refresh the cached value\nbook.save(\"mau_forecast.xlsx\")\n```\n\nThe edit example flips the **input** (`D2`), not the result — that is\nthe whole point of formula-first. If `E2` had been written as the\nliteral `=148*27*0.85`, this edit would have left ARR stale.\n`data_only=True` is **read-only safe only** — saving such a workbook\npermanently replaces every formula with its cached value. See\n[`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md) §7.\n\n### 3.3 xlsxwriter — write-only throughput\n\n```python\nimport xlsxwriter\n\nbook  = xlsxwriter.Workbook(\"big_table.xlsx\")\nsheet = book.add_worksheet(\"Tokens\")\nfor r, row in enumerate(rows):\n    sheet.write_row(r, 0, row)\nsheet.write_formula(0, 5, \"=SUM(F2:F100001)\")\nbook.close()\n```\n\nFaster than openpyxl on six-figure-row writes; cannot reopen what it\nwrote. Reach for it when the workbook is the terminal output.\n\n### 3.4 polars — fast read for huge files\n\n```python\nimport polars as pl\n\nframe    = pl.read_excel(\"big.xlsx\")\nas_pandas = frame.to_pandas()                                            # interop\n```\n\n> ⚠️ **`polars.write_excel()` writes static numbers — same trap as\n> `pandas.to_excel`.** No formulas, no charts, no styles. Polars is a\n> read-side accelerator for files above ~500k rows; for the write side,\n> always pair with openpyxl and emit derived values as `=` formulas\n> (`docs\u002Fadvanced-reference.md` §6 has the worked large-file pattern).\n\n### 3.5 pyexcel — format-agnostic conversion\n\n```python\nimport pyexcel as pe\n\nrecords = pe.get_records(file_name=\"upload.xlsx\")                       # also csv \u002F ods \u002F tsv\npe.save_book_as(file_name=\"raw.csv\", dest_file_name=\"clean.xlsx\")\n```\n\nUse it when the input format is unknown ahead of time, or when the\npipeline accepts several formats.\n\n---\n\n## 4. Recalculation Routes\n\nopenpyxl writes formulas as strings and never evaluates them.\nRecalculation is mandatory before delivery; it refreshes the cached\nvalues and surfaces the seven Excel error markers.\n\n### 4.1 `scripts\u002Frecalc.py` — LibreOffice headless (default)\n\n```bash\npython scripts\u002Frecalc.py mau_forecast.xlsx 30\n```\n\nSample success output on stdout:\n\n```json\n{\n  \"status\": \"success\",\n  \"total_errors\": 0,\n  \"total_formulas\": 42\n}\n```\n\nIf `status == \"errors_found\"`, the JSON includes an `error_summary`\nkeyed by error marker with up to 20 locations per marker. For the full\nJSON schema, the seven error markers, the macro install path, the\ncross-platform timeout wrapper, and the AF_UNIX sandbox shim, see\n[`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md).\n\n### 4.2 `xlcalculator` \u002F `formulas` — pure-Python fallback\n\nIn CI containers or serverless runtimes without LibreOffice, fall back\nto [`xlcalculator`](https:\u002F\u002Fgithub.com\u002Fbradbase\u002Fxlcalculator) or\n[`formulas`](https:\u002F\u002Fgithub.com\u002Fvinci1it2000\u002Fformulas). Neither\nsupports array formulas, pivot tables, or custom functions — for\ndelivery, always rerun `recalc.py` on a host with LibreOffice. Limits\nin [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §8.\n\n---\n\n## 5. Conventions for Excel Outputs\n\n**Formula-first as the unifying rule.** All four sub-sections below\nexist to enforce one principle: a delivered workbook is a _live model_\nthe user can adjust by editing inputs and watching every dependent\ncell recompute. The colour code (§5.2 — black for formulas, blue for\nhardcoded inputs) makes it obvious at a glance whether a number is\nderived or assumed; the recalc gate (§5.1) makes sure every formula\nactually evaluates; the number formats (§5.3) make sure the recomputed\nvalues stay readable. **Whenever you can express a value as a formula\nreferencing other cells, do so** — never paste a pre-computed total \u002F\naverage \u002F ratio \u002F growth rate as a literal number. Hard-coded numbers\nbelong only in the leading Assumptions block (full rationale in\n[`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §4).\n\n### 5.1 Professional font and zero formula errors\n\nDefault body font is one of Calibri \u002F Arial \u002F Times New Roman, used\nconsistently throughout the workbook. Every workbook with formulas\nmust pass `scripts\u002Frecalc.py` with `total_errors == 0` before delivery.\n\n### 5.2 Color coding\n\n| Color         | Hex      | RGB                    | Meaning                                    | Anti-pattern                                                     |\n| ------------- | -------- | ---------------------- | ------------------------------------------ | ---------------------------------------------------------------- |\n| Black         | `000000` | `0,0,0`                | Every formula and computed result          | Setting `=A1+B1` to blue makes it look like a manual input       |\n| Blue          | `0000FF` | `0,0,255`              | Hard-coded inputs and scenario assumptions | Setting `=Sheet2!A1` to blue hides the cross-sheet link          |\n| Green         | `008000` | `0,128,0`              | Same-workbook cross-sheet links            | Coloring a cross-file link green hides the external dependency   |\n| Red           | `FF0000` | `255,0,0`              | Cross-file external links                  | Coloring same-sheet `=A1` red implies a fake external dependency |\n| Yellow (fill) | `FFFF00` | `255,255,0` background | Critical assumptions awaiting review       | Coloring every cell yellow drowns out the cells that need review |\n\nBlack-first ordering matches the reviewer's scan order — see\n[`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §1 for the\nrationale. Three- and six-digit hex are equivalent (`#000` ≡ `#000000`);\nboth `openpyxl` and `xlsxwriter` accept either form.\n\n### 5.3 Number formatting\n\n| Type       | Format                                                               | Rationale                                                 |\n| ---------- | -------------------------------------------------------------------- | --------------------------------------------------------- |\n| Year       | text (`\"FY2025\"`)                                                    | Numeric `2025` gets thousands-separated to `2,025`        |\n| Currency   | `$#,##0` or `¥#,##0` with the unit (`mm` \u002F `bn` \u002F `k`) in the header | The unit lives once per column, not per cell              |\n| Zero       | `\"$#,##0;($#,##0);-\"`                                                | Em-dash separates \"zero by intent\" from \"missing value\"   |\n| Percentage | `0.0%`                                                               | One decimal is enough for a scan; two is false precision  |\n| Multiple   | `0.0x`                                                               | Used for valuation multiples (EV\u002FEBITDA, P\u002FS, P\u002FE)        |\n| Negative   | `(123)` (parentheses)                                                | Accounting convention; minus signs blend into table grids |\n| Hardcode   | `Source: …` annotation in the next column or as a cell comment       | Reviewer can audit every input back to its source         |\n\nSample headers: `ARR (¥mm)` \u002F `Tokens (bn)` \u002F `MAU (mm)` \u002F\n`Token unit price 1.2x`. Full hardcode-source grammar and five worked\nexamples in [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §3.\n\n### 5.4 Preserve existing templates\n\nWhen updating someone else's template, observe before editing — colors,\ncolumn widths, fonts, merged cells, number formats, and conditional\nformatting from the existing file all override this skill's defaults.\nChecklist in [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §5.\n\n---\n\n## 6. Reference Index\n\n| File                                                                                               | Purpose                                                                                                                                                                     |\n| -------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| [`docs\u002Fpitfalls-index.md`](docs\u002Fpitfalls-index.md)                                                 | **Read first when the task matches a known case.** 8 evaluation cases keyed by task signature → past failure → recommended trace                                            |\n| [`docs\u002Fsuperstore-multiformat-conversion-case.md`](docs\u002Fsuperstore-multiformat-conversion-case.md) | X8 progressive case: Superstore Excel → CSV\u002FJSON\u002FHTML→PDF + XML-template CSV→XLSX reverse validation                                                                        |\n| [`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md)                                           | openpyxl create \u002F edit \u002F chart recipes plus merged-cell and shared-strings gotchas                                                                                          |\n| [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md)                                                     | `recalc.py` reference — macro path, JSON schema, AF_UNIX shim, timeouts, alternatives                                                                                       |\n| [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md)                                           | Color coding, number formats, formula construction, hardcode source documentation                                                                                           |\n| [`docs\u002Fraw-xml-escape-hatch.md`](docs\u002Fraw-xml-escape-hatch.md)                                     | Surgical XML escape hatch via `scripts\u002Foffice\u002F{pack,unpack}.py` for the few files openpyxl can't safely round-trip (VBA, pivot caches, slicers, connections, externalLinks) |\n| [`docs\u002Fadvanced-reference.md`](docs\u002Fadvanced-reference.md)                                         | polars \u002F duckdb \u002F xlcalculator \u002F pyexcel \u002F extract-text \u002F large-file \u002F CI workflows                                                                                         |\n| `scripts\u002Frecalc.py`                                                                                | LibreOffice-driven recalculation entry point — mandatory final step                                                                                                         |\n| `scripts\u002Foffice\u002Fsoffice.py`                                                                        | Sandbox-friendly LibreOffice runner with AF_UNIX shim                                                                                                                       |\n| `scripts\u002Foffice\u002F{pack,unpack,validate}.py`                                                         | DOCX \u002F PPTX \u002F XLSX zip helpers + XSD validate (cross-format; schemas live under `office\u002Fschemas\u002F`)                                                                          |\n\n---\n\n## 7. Troubleshooting\n\n| Symptom                                                                          | Likely cause                                                                                         | First thing to try                                                                                                                                                                                                                                                                                                                                                                                         |\n| -------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| Recalc returns `#REF!` after row insertion                                       | Old formulas point at the row that just shifted                                                      | Search the workbook for `#REF!`; replace with the new coordinate; rerun `recalc.py`                                                                                                                                                                                                                                                                                                                        |\n| Recalc returns `#DIV\u002F0!`                                                         | Denominator evaluates to zero or empty                                                               | Wrap with `IFERROR(num\u002Fdenom, 0)` or `IF(denom=0, \"\", num\u002Fdenom)`                                                                                                                                                                                                                                                                                                                                          |\n| Recalc returns `#N\u002FA`                                                            | `VLOOKUP` \u002F `MATCH` key not found in the lookup column                                               | Check key whitespace, case, and type (`123` vs `\"123\"`); confirm the lookup column is contiguous                                                                                                                                                                                                                                                                                                           |\n| `recalc.py` exits with `soffice timed out`                                       | LibreOffice hung on a heavy workbook                                                                 | Raise the second positional arg (`python scripts\u002Frecalc.py file.xlsx 180`); if it persists, see `docs\u002Frecalc-guide.md` §5                                                                                                                                                                                                                                                                                  |\n| `recalc.py` errors with `Address already in use` \u002F AF_UNIX denied                | Sandbox blocks AF_UNIX (macOS App Sandbox \u002F Linux seccomp)                                           | Auto-handled by `office.soffice` shim; ensure `gcc` is on `PATH` so the shim can compile (see `docs\u002Frecalc-guide.md` §6)                                                                                                                                                                                                                                                                                   |\n| `iter_rows()` returns `None` for cells that visibly have a value                 | Cells live inside a merged region; only the anchor carries the value                                 | `unmerge_cells()` first, broadcast the anchor value, then re-merge if layout matters (see `docs\u002Fcreate-edit-guide.md` §9)                                                                                                                                                                                                                                                                                  |\n| `save()` deletes every formula in the workbook                                   | The file was opened with `data_only=True`                                                            | Reload with `load_workbook(path)` (default `data_only=False`); `data_only=True` is read-only safe only                                                                                                                                                                                                                                                                                                     |\n| `load_workbook` raises `TypeError: expected \u003Cclass 'int'>` after `recalc.py` ran | LibreOffice rewrote the workbook's `\u003CmergeCell>` XML in a form openpyxl's default parser cannot read | The file itself is intact and openable in Excel — `recalc.py` already returned `status: success` via its raw-XML fallback (see `docs\u002Frecalc-guide.md` §9). For programmatic readback, retry with `load_workbook(path, read_only=True, data_only=True)` (streaming reader skips the merged-cell parser); when you also need formula strings, drop to the raw-XML pattern in `docs\u002Fcreate-edit-guide.md` §11 |\n| Polars `write_excel` output has no formulas, no charts, no styles                | Expected — polars does not write formulas \u002F charts \u002F styles                                          | Use openpyxl on the write side (see `docs\u002Fadvanced-reference.md` §6)                                                                                                                                                                                                                                                                                                                                       |\n| `load_workbook` strips VBA \u002F pivot tables \u002F slicers \u002F connections                | openpyxl can't round-trip these binary or strongly-coupled artefacts                                 | Drop to `scripts\u002Foffice\u002Funpack.py` + manual XML edit + `scripts\u002Foffice\u002Fpack.py`; see `docs\u002Fraw-xml-escape-hatch.md` §1 for the full no-go table                                                                                                                                                                                                                                                            |\n\n---\n\n## 8. Environment\n\n| Dependency                      | Purpose                                                  | Install                                                                            |\n| ------------------------------- | -------------------------------------------------------- | ---------------------------------------------------------------------------------- |\n| Python 3.9+                     | All `.py` scripts                                        | system                                                                             |\n| `openpyxl`                      | Read \u002F write `.xlsx` with formulas, styles, named ranges | `pip install openpyxl`                                                             |\n| `pandas`                        | Tabular read \u002F write (default for §3.1)                  | `pip install pandas openpyxl`                                                      |\n| `polars` (optional)             | Fast read for large files (§3.4)                         | `pip install polars`                                                               |\n| `xlsxwriter` (optional)         | Write-only throughput (§3.3)                             | `pip install xlsxwriter`                                                           |\n| `pyexcel` (optional)            | Format-agnostic conversion (§3.5)                        | `pip install pyexcel pyexcel-xlsx`                                                 |\n| LibreOffice (`soffice` on PATH) | `recalc.py` headless recalculation                       | macOS: `brew install --cask libreoffice`; Debian\u002FUbuntu: `apt install libreoffice` |\n| `coreutils` (macOS)             | Provides `gtimeout` for the recalc timeout wrapper       | `brew install coreutils` (optional; falls back to `subprocess` kwarg)              |\n\n`scripts\u002Foffice\u002Fsoffice.py` auto-activates an `LD_PRELOAD` AF_UNIX shim\nwhen the host sandbox denies AF_UNIX (macOS App Sandbox, Linux seccomp);\n`recalc.py` already routes through `get_soffice_env()`, so it is transparent.\n\nFor long-running batch jobs (hundreds of workbooks per hour), swap\n`soffice` per-call for [`unoserver`](https:\u002F\u002Fgithub.com\u002Funoconv\u002Funoserver) —\nsee [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §7.\n\n## Windows (win32) platform notes\n\nScripts and code examples in this skill use bash syntax (heredocs, pipes, `python3`). On Windows,\nexecute them via **Git Bash**:\n\n```powershell\n# Raw XML escape hatch\nbash -c \"python scripts\u002Foffice\u002Funpack.py input.xlsx \u002Ftmp\u002Fwork\u002F\"\nbash -c \"python scripts\u002Foffice\u002Fpack.py \u002Ftmp\u002Fwork\u002F output.xlsx\"\n\n# Piped inspection\nbash -c \"python -c \\\"from openpyxl import load_workbook; wb=load_workbook('file.xlsx', read_only=True); print(wb.sheetnames)\\\"\"\n```\n\n> **⚠️ recalc.py limitation on Windows:** `scripts\u002Frecalc.py` installs a LibreOffice Basic macro\n> to `~\u002F.config\u002Flibreoffice\u002F...` (Linux path). On Windows, LibreOffice stores macros in\n> `%APPDATA%\\LibreOffice\\4\\user\\basic\\Standard\\`. Until `recalc.py` is patched to resolve the\n> Windows macro directory, recalc will fail to find the macro. **Workaround:** open the workbook\n> in LibreOffice GUI, run Ctrl+Shift+F9 (recalculate all), then save.\n\n**Key differences on Windows:**\n\n| Unix reference                    | Windows handling                                                  |\n| --------------------------------- | ----------------------------------------------------------------- |\n| `python3`                         | Use `python` — Git Bash finds it on PATH if installed             |\n| `python3 \u003C\u003CEOF ... EOF`           | Use `bash -c \"python -c '...'\"` or write a .py file and run it    |\n| `\u002Ftmp\u002F`                           | Git Bash maps `\u002Ftmp\u002F` automatically — scripts work as-is          |\n| `grep -in ... \\| head -20`        | Works inside `bash -c \"...\"`                                      |\n| `brew install --cask libreoffice` | `winget install TheDocumentFoundation.LibreOffice`                |\n| `coreutils` (`gtimeout`)          | Not needed — `recalc.py` falls back to `subprocess` timeout kwarg |\n\n**If Git Bash or any tool is missing**, detect before installing anything:\n\n| Tool        | Detect                                                     | Install (run by the user, not the agent)           |\n| ----------- | ---------------------------------------------------------- | -------------------------------------------------- |\n| Python      | `python --version`                                         | `winget install Python.Python.3.12`                |\n| Git Bash    | `ls \"$PROGRAMFILES\u002FGit\u002Fbin\u002Fbash.exe\"`                      | `winget install Git.Git`                           |\n| LibreOffice | `ls \"$PROGRAMFILES\u002FLibreOffice\u002Fprogram\u002Fsoffice.exe\"`       | `winget install TheDocumentFoundation.LibreOffice` |\n| openpyxl    | `python -c \"import openpyxl; print(openpyxl.__version__)\"` | `python -m pip install openpyxl`                   |\n\nThis fork deliberately does **not** ship an auto-installer. Upstream delegated the bootstrap to a\nreference inside a different skill in its own host product; that skill is not part of ZCodium, so\nthe pointer dangles. A skill that also writes to the user's shell config or installs global\npackages is the wrong default on a managed host anyway. Report what is missing and let the user\nrun the install command.\n","\u003C!--\nCopied from MiniMaxAI\u002Fminimax-code (packages\u002Flocal-runtime\u002Fassets\u002Fskills\u002Fxlsx)\nat revision 564e9166d81f87b0b767b005e4779d4697b512be.\nCopyright (c) 2026 MiniMax Code. Licensed under MIT.\nModified by ZCode: added the provenance notice; replaced a pointer into a different skill in the upstream host product with a self-contained Windows tool table (detect + install command), and stated that this fork ships no auto-installer. All other skill instructions unchanged.\nSee THIRD-PARTY-NOTICES.md in the repository root for license and provenance.\n-->\n\n# xlsx\n\nA pragmatic, recipe-first guide for reading, editing, creating, and\nrecalculating `.xlsx` \u002F `.xlsm` \u002F `.csv` \u002F `.tsv` files. The default\npairing is **pandas for tabular data + openpyxl for formulas, styles,\nnamed ranges, and charts**. Recalculation through\n[`scripts\u002Frecalc.py`](scripts\u002Frecalc.py) (LibreOffice headless) is\nmandatory before delivery — openpyxl writes formulas as strings and\nnever evaluates them.\n\n> **Formula-first.** A spreadsheet without live formulas is just a\n> CSV with a fancier extension. **Every computed value** — totals,\n> averages, growth rates, ratios, cross-sheet references, percent\n> changes, anything derivable from other cells — **must be written as\n> a live `=…` formula**, not as the pre-computed number. Hard-coded\n> numbers belong only in the **Assumptions** block (inputs the user\n> can flip to re-run the model). When in doubt, write the formula.\n> Full convention in §5 and\n> [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §4.\n>\n> **❌ The most common anti-pattern.** Loading a workbook with pandas,\n> computing derived columns in Python (`df[\"total\"] = df[\"a\"] + df[\"b\"]`,\n> `df.groupby(...).sum()`, etc.), then writing the result back with\n> `df.to_excel(...)` ships **static numbers** — the workbook becomes a\n> dead snapshot the moment any input changes. Use pandas\u002Fpolars only\n> to load and clean **raw inputs**; emit every derived value as a live\n> `=` formula via openpyxl (§3.2). This rule applies regardless of how\n> easy it would be to compute the value in Python first.\n\n> **Spreadsheet output only.** For Word documents, PowerPoint slides,\n> HTML reports, standalone Python scripts, database pipelines, or the\n> Google Sheets API, switch to the matching skill.\n\n## Operational rules — read before doing anything\n\n> **1. Match user query against [`docs\u002Fpitfalls-index.md`](docs\u002Fpitfalls-index.md)\n> FIRST.** It contains 8 production-ready **canonical query templates**\n> (X1–X8), each with a `Match signatures` block (sample queries) and a\n> complete executable prompt that already encodes Formula-first, recalc\n> verification, formula-count gate, sample-data-integrity, slicer\n> handling, and every other rule below. Workflow:\n>\n> 1. Scan the Quick lookup table — match user's query keywords to a row.\n> 2. **Copy the matching canonical query verbatim**, substitute the\n>    `Slots` (e.g. `{INPUT}`, `{COMPANY}`, `{OUTPUT_XLSX}`) with the\n>    user's actual values, and execute step-by-step.\n> 3. Multiple partial matches → fuse: take the strictest verification\n>    from each, never relax a constraint.\n> 4. No match → fall back to the Decision Tree \u002F per-section guides below.\n>\n> Do NOT skip verification steps in the canonical queries — every\n> \"ship a wrong workbook\" failure traces back to skipping recalc,\n> total_formulas check, or row-count canary.\n\n> **2. Formula-first is non-negotiable.** A delivery that ships static\n> numbers where formulas were possible is a **failed delivery** — even\n> if every number is numerically correct the moment you saved it. The\n> user's first edit will expose the lie. Hardcode only the Assumptions\n> block; everything derivable from other cells goes in as `=…`. Full\n> regime in §5.\n\n> **3. `total_formulas == 0` is a red flag, not a green light.** A\n> \"success\" recalc with zero formulas means the workbook is a static\n> dump — the user can't audit, can't re-run scenarios, can't\n> introspect derivations. Treat it as a delivery failure and rewrite\n> via §3.2 \u002F openpyxl `=` formulas. The only legitimate exception is\n> a workbook the user explicitly asked to be a static snapshot\n> (in which case `pandas.to_excel` is the right tool, not this skill).\n\n> **4. Source-data-integrity rule — never `df.sample(N)` \u002F `df.head(N)`\n> on the raw sheet.** Down-sampling 400k rows to 100k \"because openpyxl\n> writes are slow\" silently destroys every aggregation built on top —\n> the pivot or `=SUMIF` summary will be off by ~75% and look entirely\n> plausible. **Write the full row count, even if it takes minutes.**\n> If write throughput is the real bottleneck, switch the writer\n> (xlsxwriter §3.3, or `Workbook(write_only=True)` streaming pattern\n> in [`docs\u002Fadvanced-reference.md`](docs\u002Fadvanced-reference.md) §6),\n> never the sample size.\n\n> **5. Summaries must be Excel-native — pivot tables or `=SUMIFS` \u002F\n> `=COUNTIFS` over the Raw sheet, never Python `groupby` written back\n> as values.** This is the same rule as §3 phrased for the most\n> common offender. The user expects: edit a Raw row, hit recalc, see\n> the totals move. Static summaries break that contract silently.\n\n> **6. Slicers cannot be authored from scratch by openpyxl.** The\n> `xl\u002Fslicers\u002F*.xml` part requires GUID-bound pivot cache references\n> that openpyxl has no API for. Two production paths:\n> (a) template-inheritance — author a `template.xlsx` once in Excel\n> or LibreOffice with the slicer wired to a named range, then\n> `load_workbook(template) → write into named range → save as new`;\n> (b) raw-XML transplant from a known-good template via\n> `scripts\u002Foffice\u002F{unpack,pack}.py`. **If the user asks for a slicer\n> and you have no template, say so before silently downgrading to a\n> static dropdown.**\n\n> **7. Don't suppress stderr.** `2>\u002Fdev\u002Fnull` is **never** the right\n> choice in this skill (`recalc.py`, `soffice`, `pdfinfo`, `unpack.py`,\n> `pack.py`). On failure you lose the only signal that explains why\n> and have to rerun blind. If output is too noisy, redirect to a log\n> file and grep on demand:\n>\n> ```bash\n> python scripts\u002Frecalc.py file.xlsx 60 2>\u002Ftmp\u002Frecalc.log\n> # If JSON shows status != \"success\" or exit != 0, then:\n> #   grep -in \"error\\|trace\\|fail\" \u002Ftmp\u002Frecalc.log | head -20\n> ```\n\n> **8. When the user states a numeric range for a derived value,\n> enforce it in the formula.** \"Composite score 0–100\" → wrap the\n> formula with `=ROUND(MIN(MAX(raw, 0), 100), 1)`. Don't ship a\n> 0–1804 column and blame \"data anomaly\". Spot-check `MIN()` \u002F\n> `MAX()` of the output column after recalc.\n\n> **9. 500k+ rows: pandas\u002Fpolars first, not openpyxl row walking.**\n> For large tabular inputs (roughly **500,000+ rows**, or hundreds of MB),\n> default to `pandas`\u002F`polars` for reading, filtering, type normalization,\n> joins, and QA spot-checks. Do **not** iterate cell-by-cell with openpyxl to\n> inspect or transform the source workbook — it is too slow and encourages\n> accidental sampling. Use openpyxl only at the output boundary for formulas,\n> styles, charts, templates, and final `.xlsx` assembly. If the output is a\n> static raw-data dump, `DataFrame.to_excel`\u002F`xlsxwriter` is acceptable; if it\n> contains derived values, write full raw rows and Excel-native `=` formulas.\n\n> **10. Non-standard XLSX packages: trust recalc raw-XML fallback.**\n> Workbooks with heavy merged cells, charts\u002Fdrawings, or vendor-generated XML\n> can make openpyxl crash even after LibreOffice recalculated successfully.\n> `scripts\u002Frecalc.py` now catches openpyxl parse failures and falls back to a\n> raw worksheet XML scanner. If JSON returns `scanner: \"raw_xml_fallback\"` with\n> `status: success` \u002F `errors_found`, treat it as authoritative and do not\n> retry blindly. Use the `compatibility_hint` field to decide whether to avoid\n> downstream openpyxl rewrites. Details in [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §10.\n\nFor deeper material — creation and edit recipes, recalc internals,\nfinancial-model conventions, large-file alternatives — see\n[`docs\u002F`](docs\u002F).\n\n---\n\n## 1. Scope and When to Use\n\nThis skill covers the everyday spreadsheet chores an LLM agent is most\noften asked to perform end-to-end:\n\n- inspecting an existing workbook (sheet names, merged regions, headers)\n- reading tabular data into pandas \u002F polars for analysis or QA\n- creating a new workbook with formulas, styles, named ranges, and charts\n- editing an existing workbook in place (insert \u002F delete \u002F restyle \u002F replace)\n- recalculating every formula via [`scripts\u002Frecalc.py`](scripts\u002Frecalc.py)\n  and verifying `total_errors == 0` in the JSON\n- enforcing the financial-model conventions in §5 on every delivery\n- converting between tabular formats (`csv` ↔ `xlsx` ↔ `ods` ↔ `tsv`)\n\nUse a different approach when:\n\n- the deliverable is a Word document, slide deck, standalone script, or\n  database pipeline — switch to the matching skill (`docx`, `pptx`, etc.)\n- the workflow is read-only and the user just wants the cell values\n  printed — `extract-text` (`docs\u002Fadvanced-reference.md` §5) is enough\n- the workbook lives behind the Google Sheets API rather than on disk\n\n---\n\n## 2. Decision Tree\n\n```\nWhat does the user want from \u002F for this workbook?\n|\n+-- Read tabular data only ───────────────> pandas \u002F polars     -> §3.1, §3.4\n+-- Inspect quickly without code ─────────> extract-text CLI    -> docs\u002Fadvanced-reference.md §5\n+-- Create a new workbook ────────────────> openpyxl            -> §3.2\n+-- Edit an existing workbook in place ───> openpyxl            -> §3.2\n+-- Output has any derived\u002Fcomputed cells > openpyxl + `=` formulas -> §3.2, §5\n+-- Very large workbook (500k+ rows) ─────> pandas\u002Fpolars read + openpyxl\u002Fxlsxwriter output -> §3.1, §3.4, docs\u002Fadvanced-reference.md §6\n+-- Recalculate formulas (mandatory) ─────> scripts\u002Frecalc.py   -> §4.1\n+-- Convert format (csv ↔ xlsx ↔ ods) ────> pyexcel             -> §3.5\n+-- Sandboxed env (no AF_UNIX) ───────────> office.soffice shim -> §4.1\n```\n\n**Default rule.** Read with pandas, write and edit with openpyxl,\nrecalculate with `scripts\u002Frecalc.py`. Drop to xlsxwriter when the\nworkload is write-only and throughput-bound; drop to polars when the\ninput file is too big for pandas to hold in memory. **Never use\n`pandas.to_excel` \u002F `polars.write_excel` for derived\u002Fcomputed values\n— those write static numbers, not live formulas. See §3.1's ⚠️ box.**\n\n---\n\n## 3. Library Cookbook\n\nFive subsections, one per library. Each entry is a minimum viable\nexample plus a short note. Deeper recipes — full styling, conditional\nformatting, charts, edit gotchas — live in\n[`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md).\n\n### 3.1 pandas — tabular read \u002F write (default)\n\n```python\nimport pandas as pd\n\nframe = pd.read_excel(\"mau_forecast.xlsx\", engine=\"openpyxl\")           # first sheet\nsheets = pd.read_excel(\"mau_forecast.xlsx\", sheet_name=None)            # {sheet: DataFrame}\nframe.to_excel(\"clean_mau.xlsx\", sheet_name=\"clean\", index=False)\n```\n\n> ⚠️ **`to_excel()` writes static numbers only — no formulas survive.**\n> Use it strictly for **raw data dumps**: cleaned input data, exported\n> query results, anything where every cell is itself a primary value.\n> The moment the deliverable contains a derived value — totals,\n> averages, ratios, growth rates, anything computable from other cells\n> — switch to §3.2 and write it as a live `=` formula. **Never** do\n> `df[\"total\"] = df[\"a\"] + df[\"b\"]; df.to_excel(...)`: the workbook\n> will look fine until the user edits an input and discovers the\n> \"totals\" don't move. For aggregations across rows\u002Fsheets, write the\n> raw rows here and emit `=SUM(...)` \u002F `=SUMIF(...)` on the openpyxl\n> side (`docs\u002Fadvanced-reference.md` §6 has the worked large-file\n> example).\n\nDefault for any \"read tabular data\" job. For 500k+ rows, prefer pandas\u002Fpolars\nfor source inspection, filtering, joins, and QA; avoid openpyxl cell-by-cell\nsource traversal. Pair with openpyxl only on the write side when you need\nformulas or formatting.\n\n### 3.2 openpyxl — create and edit (default for write + edit)\n\n```python\nfrom openpyxl import Workbook, load_workbook\nfrom openpyxl.styles import Font\n\nbook  = Workbook()                                                       # create\nsheet = book.active\nsheet[\"A1\"], sheet[\"B1\"], sheet[\"C1\"], sheet[\"D1\"], sheet[\"E1\"] = (\n    \"Quarter\", \"MAU (mm)\", \"Tokens\u002FMAU\", \"Take-rate\", \"ARR (¥mm)\"\n)\n# Inputs (Assumptions) — hardcoded primary values; §5.2 requires blue.\ninput_font = Font(color=\"0000FF\")\nsheet[\"A2\"], sheet[\"B2\"], sheet[\"C2\"], sheet[\"D2\"] = \"2025Q3\", 148, 27, 0.85\nfor ref in (\"B2\", \"C2\", \"D2\"):\n    sheet[ref].font = input_font\n# Derived ARR — references inputs, not literal numbers; §5.2 requires black.\nsheet[\"E2\"] = \"=B2*C2*D2\"\nsheet[\"E2\"].font = Font(color=\"000000\")\nbook.save(\"mau_forecast.xlsx\")\n\nbook = load_workbook(\"mau_forecast.xlsx\")                               # edit\nbook[\"Sheet\"][\"D2\"] = 0.90                                              # flip take-rate\n# E2 still says \"=B2*C2*D2\" — recalc.py will refresh the cached value\nbook.save(\"mau_forecast.xlsx\")\n```\n\nThe edit example flips the **input** (`D2`), not the result — that is\nthe whole point of formula-first. If `E2` had been written as the\nliteral `=148*27*0.85`, this edit would have left ARR stale.\n`data_only=True` is **read-only safe only** — saving such a workbook\npermanently replaces every formula with its cached value. See\n[`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md) §7.\n\n### 3.3 xlsxwriter — write-only throughput\n\n```python\nimport xlsxwriter\n\nbook  = xlsxwriter.Workbook(\"big_table.xlsx\")\nsheet = book.add_worksheet(\"Tokens\")\nfor r, row in enumerate(rows):\n    sheet.write_row(r, 0, row)\nsheet.write_formula(0, 5, \"=SUM(F2:F100001)\")\nbook.close()\n```\n\nFaster than openpyxl on six-figure-row writes; cannot reopen what it\nwrote. Reach for it when the workbook is the terminal output.\n\n### 3.4 polars — fast read for huge files\n\n```python\nimport polars as pl\n\nframe    = pl.read_excel(\"big.xlsx\")\nas_pandas = frame.to_pandas()                                            # interop\n```\n\n> ⚠️ **`polars.write_excel()` writes static numbers — same trap as\n> `pandas.to_excel`.** No formulas, no charts, no styles. Polars is a\n> read-side accelerator for files above ~500k rows; for the write side,\n> always pair with openpyxl and emit derived values as `=` formulas\n> (`docs\u002Fadvanced-reference.md` §6 has the worked large-file pattern).\n\n### 3.5 pyexcel — format-agnostic conversion\n\n```python\nimport pyexcel as pe\n\nrecords = pe.get_records(file_name=\"upload.xlsx\")                       # also csv \u002F ods \u002F tsv\npe.save_book_as(file_name=\"raw.csv\", dest_file_name=\"clean.xlsx\")\n```\n\nUse it when the input format is unknown ahead of time, or when the\npipeline accepts several formats.\n\n---\n\n## 4. Recalculation Routes\n\nopenpyxl writes formulas as strings and never evaluates them.\nRecalculation is mandatory before delivery; it refreshes the cached\nvalues and surfaces the seven Excel error markers.\n\n### 4.1 `scripts\u002Frecalc.py` — LibreOffice headless (default)\n\n```bash\npython scripts\u002Frecalc.py mau_forecast.xlsx 30\n```\n\nSample success output on stdout:\n\n```json\n{\n  \"status\": \"success\",\n  \"total_errors\": 0,\n  \"total_formulas\": 42\n}\n```\n\nIf `status == \"errors_found\"`, the JSON includes an `error_summary`\nkeyed by error marker with up to 20 locations per marker. For the full\nJSON schema, the seven error markers, the macro install path, the\ncross-platform timeout wrapper, and the AF_UNIX sandbox shim, see\n[`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md).\n\n### 4.2 `xlcalculator` \u002F `formulas` — pure-Python fallback\n\nIn CI containers or serverless runtimes without LibreOffice, fall back\nto [`xlcalculator`](https:\u002F\u002Fgithub.com\u002Fbradbase\u002Fxlcalculator) or\n[`formulas`](https:\u002F\u002Fgithub.com\u002Fvinci1it2000\u002Fformulas). Neither\nsupports array formulas, pivot tables, or custom functions — for\ndelivery, always rerun `recalc.py` on a host with LibreOffice. Limits\nin [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §8.\n\n---\n\n## 5. Conventions for Excel Outputs\n\n**Formula-first as the unifying rule.** All four sub-sections below\nexist to enforce one principle: a delivered workbook is a _live model_\nthe user can adjust by editing inputs and watching every dependent\ncell recompute. The colour code (§5.2 — black for formulas, blue for\nhardcoded inputs) makes it obvious at a glance whether a number is\nderived or assumed; the recalc gate (§5.1) makes sure every formula\nactually evaluates; the number formats (§5.3) make sure the recomputed\nvalues stay readable. **Whenever you can express a value as a formula\nreferencing other cells, do so** — never paste a pre-computed total \u002F\naverage \u002F ratio \u002F growth rate as a literal number. Hard-coded numbers\nbelong only in the leading Assumptions block (full rationale in\n[`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §4).\n\n### 5.1 Professional font and zero formula errors\n\nDefault body font is one of Calibri \u002F Arial \u002F Times New Roman, used\nconsistently throughout the workbook. Every workbook with formulas\nmust pass `scripts\u002Frecalc.py` with `total_errors == 0` before delivery.\n\n### 5.2 Color coding\n\n| Color         | Hex      | RGB                    | Meaning                                    | Anti-pattern                                                     |\n| ------------- | -------- | ---------------------- | ------------------------------------------ | ---------------------------------------------------------------- |\n| Black         | `000000` | `0,0,0`                | Every formula and computed result          | Setting `=A1+B1` to blue makes it look like a manual input       |\n| Blue          | `0000FF` | `0,0,255`              | Hard-coded inputs and scenario assumptions | Setting `=Sheet2!A1` to blue hides the cross-sheet link          |\n| Green         | `008000` | `0,128,0`              | Same-workbook cross-sheet links            | Coloring a cross-file link green hides the external dependency   |\n| Red           | `FF0000` | `255,0,0`              | Cross-file external links                  | Coloring same-sheet `=A1` red implies a fake external dependency |\n| Yellow (fill) | `FFFF00` | `255,255,0` background | Critical assumptions awaiting review       | Coloring every cell yellow drowns out the cells that need review |\n\nBlack-first ordering matches the reviewer's scan order — see\n[`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §1 for the\nrationale. Three- and six-digit hex are equivalent (`#000` ≡ `#000000`);\nboth `openpyxl` and `xlsxwriter` accept either form.\n\n### 5.3 Number formatting\n\n| Type       | Format                                                               | Rationale                                                 |\n| ---------- | -------------------------------------------------------------------- | --------------------------------------------------------- |\n| Year       | text (`\"FY2025\"`)                                                    | Numeric `2025` gets thousands-separated to `2,025`        |\n| Currency   | `$#,##0` or `¥#,##0` with the unit (`mm` \u002F `bn` \u002F `k`) in the header | The unit lives once per column, not per cell              |\n| Zero       | `\"$#,##0;($#,##0);-\"`                                                | Em-dash separates \"zero by intent\" from \"missing value\"   |\n| Percentage | `0.0%`                                                               | One decimal is enough for a scan; two is false precision  |\n| Multiple   | `0.0x`                                                               | Used for valuation multiples (EV\u002FEBITDA, P\u002FS, P\u002FE)        |\n| Negative   | `(123)` (parentheses)                                                | Accounting convention; minus signs blend into table grids |\n| Hardcode   | `Source: …` annotation in the next column or as a cell comment       | Reviewer can audit every input back to its source         |\n\nSample headers: `ARR (¥mm)` \u002F `Tokens (bn)` \u002F `MAU (mm)` \u002F\n`Token unit price 1.2x`. Full hardcode-source grammar and five worked\nexamples in [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §3.\n\n### 5.4 Preserve existing templates\n\nWhen updating someone else's template, observe before editing — colors,\ncolumn widths, fonts, merged cells, number formats, and conditional\nformatting from the existing file all override this skill's defaults.\nChecklist in [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md) §5.\n\n---\n\n## 6. Reference Index\n\n| File                                                                                               | Purpose                                                                                                                                                                     |\n| -------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| [`docs\u002Fpitfalls-index.md`](docs\u002Fpitfalls-index.md)                                                 | **Read first when the task matches a known case.** 8 evaluation cases keyed by task signature → past failure → recommended trace                                            |\n| [`docs\u002Fsuperstore-multiformat-conversion-case.md`](docs\u002Fsuperstore-multiformat-conversion-case.md) | X8 progressive case: Superstore Excel → CSV\u002FJSON\u002FHTML→PDF + XML-template CSV→XLSX reverse validation                                                                        |\n| [`docs\u002Fcreate-edit-guide.md`](docs\u002Fcreate-edit-guide.md)                                           | openpyxl create \u002F edit \u002F chart recipes plus merged-cell and shared-strings gotchas                                                                                          |\n| [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md)                                                     | `recalc.py` reference — macro path, JSON schema, AF_UNIX shim, timeouts, alternatives                                                                                       |\n| [`docs\u002Fconventions-guide.md`](docs\u002Fconventions-guide.md)                                           | Color coding, number formats, formula construction, hardcode source documentation                                                                                           |\n| [`docs\u002Fraw-xml-escape-hatch.md`](docs\u002Fraw-xml-escape-hatch.md)                                     | Surgical XML escape hatch via `scripts\u002Foffice\u002F{pack,unpack}.py` for the few files openpyxl can't safely round-trip (VBA, pivot caches, slicers, connections, externalLinks) |\n| [`docs\u002Fadvanced-reference.md`](docs\u002Fadvanced-reference.md)                                         | polars \u002F duckdb \u002F xlcalculator \u002F pyexcel \u002F extract-text \u002F large-file \u002F CI workflows                                                                                         |\n| `scripts\u002Frecalc.py`                                                                                | LibreOffice-driven recalculation entry point — mandatory final step                                                                                                         |\n| `scripts\u002Foffice\u002Fsoffice.py`                                                                        | Sandbox-friendly LibreOffice runner with AF_UNIX shim                                                                                                                       |\n| `scripts\u002Foffice\u002F{pack,unpack,validate}.py`                                                         | DOCX \u002F PPTX \u002F XLSX zip helpers + XSD validate (cross-format; schemas live under `office\u002Fschemas\u002F`)                                                                          |\n\n---\n\n## 7. Troubleshooting\n\n| Symptom                                                                          | Likely cause                                                                                         | First thing to try                                                                                                                                                                                                                                                                                                                                                                                         |\n| -------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| Recalc returns `#REF!` after row insertion                                       | Old formulas point at the row that just shifted                                                      | Search the workbook for `#REF!`; replace with the new coordinate; rerun `recalc.py`                                                                                                                                                                                                                                                                                                                        |\n| Recalc returns `#DIV\u002F0!`                                                         | Denominator evaluates to zero or empty                                                               | Wrap with `IFERROR(num\u002Fdenom, 0)` or `IF(denom=0, \"\", num\u002Fdenom)`                                                                                                                                                                                                                                                                                                                                          |\n| Recalc returns `#N\u002FA`                                                            | `VLOOKUP` \u002F `MATCH` key not found in the lookup column                                               | Check key whitespace, case, and type (`123` vs `\"123\"`); confirm the lookup column is contiguous                                                                                                                                                                                                                                                                                                           |\n| `recalc.py` exits with `soffice timed out`                                       | LibreOffice hung on a heavy workbook                                                                 | Raise the second positional arg (`python scripts\u002Frecalc.py file.xlsx 180`); if it persists, see `docs\u002Frecalc-guide.md` §5                                                                                                                                                                                                                                                                                  |\n| `recalc.py` errors with `Address already in use` \u002F AF_UNIX denied                | Sandbox blocks AF_UNIX (macOS App Sandbox \u002F Linux seccomp)                                           | Auto-handled by `office.soffice` shim; ensure `gcc` is on `PATH` so the shim can compile (see `docs\u002Frecalc-guide.md` §6)                                                                                                                                                                                                                                                                                   |\n| `iter_rows()` returns `None` for cells that visibly have a value                 | Cells live inside a merged region; only the anchor carries the value                                 | `unmerge_cells()` first, broadcast the anchor value, then re-merge if layout matters (see `docs\u002Fcreate-edit-guide.md` §9)                                                                                                                                                                                                                                                                                  |\n| `save()` deletes every formula in the workbook                                   | The file was opened with `data_only=True`                                                            | Reload with `load_workbook(path)` (default `data_only=False`); `data_only=True` is read-only safe only                                                                                                                                                                                                                                                                                                     |\n| `load_workbook` raises `TypeError: expected \u003Cclass 'int'>` after `recalc.py` ran | LibreOffice rewrote the workbook's `\u003CmergeCell>` XML in a form openpyxl's default parser cannot read | The file itself is intact and openable in Excel — `recalc.py` already returned `status: success` via its raw-XML fallback (see `docs\u002Frecalc-guide.md` §9). For programmatic readback, retry with `load_workbook(path, read_only=True, data_only=True)` (streaming reader skips the merged-cell parser); when you also need formula strings, drop to the raw-XML pattern in `docs\u002Fcreate-edit-guide.md` §11 |\n| Polars `write_excel` output has no formulas, no charts, no styles                | Expected — polars does not write formulas \u002F charts \u002F styles                                          | Use openpyxl on the write side (see `docs\u002Fadvanced-reference.md` §6)                                                                                                                                                                                                                                                                                                                                       |\n| `load_workbook` strips VBA \u002F pivot tables \u002F slicers \u002F connections                | openpyxl can't round-trip these binary or strongly-coupled artefacts                                 | Drop to `scripts\u002Foffice\u002Funpack.py` + manual XML edit + `scripts\u002Foffice\u002Fpack.py`; see `docs\u002Fraw-xml-escape-hatch.md` §1 for the full no-go table                                                                                                                                                                                                                                                            |\n\n---\n\n## 8. Environment\n\n| Dependency                      | Purpose                                                  | Install                                                                            |\n| ------------------------------- | -------------------------------------------------------- | ---------------------------------------------------------------------------------- |\n| Python 3.9+                     | All `.py` scripts                                        | system                                                                             |\n| `openpyxl`                      | Read \u002F write `.xlsx` with formulas, styles, named ranges | `pip install openpyxl`                                                             |\n| `pandas`                        | Tabular read \u002F write (default for §3.1)                  | `pip install pandas openpyxl`                                                      |\n| `polars` (optional)             | Fast read for large files (§3.4)                         | `pip install polars`                                                               |\n| `xlsxwriter` (optional)         | Write-only throughput (§3.3)                             | `pip install xlsxwriter`                                                           |\n| `pyexcel` (optional)            | Format-agnostic conversion (§3.5)                        | `pip install pyexcel pyexcel-xlsx`                                                 |\n| LibreOffice (`soffice` on PATH) | `recalc.py` headless recalculation                       | macOS: `brew install --cask libreoffice`; Debian\u002FUbuntu: `apt install libreoffice` |\n| `coreutils` (macOS)             | Provides `gtimeout` for the recalc timeout wrapper       | `brew install coreutils` (optional; falls back to `subprocess` kwarg)              |\n\n`scripts\u002Foffice\u002Fsoffice.py` auto-activates an `LD_PRELOAD` AF_UNIX shim\nwhen the host sandbox denies AF_UNIX (macOS App Sandbox, Linux seccomp);\n`recalc.py` already routes through `get_soffice_env()`, so it is transparent.\n\nFor long-running batch jobs (hundreds of workbooks per hour), swap\n`soffice` per-call for [`unoserver`](https:\u002F\u002Fgithub.com\u002Funoconv\u002Funoserver) —\nsee [`docs\u002Frecalc-guide.md`](docs\u002Frecalc-guide.md) §7.\n\n## Windows (win32) platform notes\n\nScripts and code examples in this skill use bash syntax (heredocs, pipes, `python3`). On Windows,\nexecute them via **Git Bash**:\n\n```powershell\n# Raw XML escape hatch\nbash -c \"python scripts\u002Foffice\u002Funpack.py input.xlsx \u002Ftmp\u002Fwork\u002F\"\nbash -c \"python scripts\u002Foffice\u002Fpack.py \u002Ftmp\u002Fwork\u002F output.xlsx\"\n\n# Piped inspection\nbash -c \"python -c \\\"from openpyxl import load_workbook; wb=load_workbook('file.xlsx', read_only=True); print(wb.sheetnames)\\\"\"\n```\n\n> **⚠️ recalc.py limitation on Windows:** `scripts\u002Frecalc.py` installs a LibreOffice Basic macro\n> to `~\u002F.config\u002Flibreoffice\u002F...` (Linux path). On Windows, LibreOffice stores macros in\n> `%APPDATA%\\LibreOffice\\4\\user\\basic\\Standard\\`. Until `recalc.py` is patched to resolve the\n> Windows macro directory, recalc will fail to find the macro. **Workaround:** open the workbook\n> in LibreOffice GUI, run Ctrl+Shift+F9 (recalculate all), then save.\n\n**Key differences on Windows:**\n\n| Unix reference                    | Windows handling                                                  |\n| --------------------------------- | ----------------------------------------------------------------- |\n| `python3`                         | Use `python` — Git Bash finds it on PATH if installed             |\n| `python3 \u003C\u003CEOF ... EOF`           | Use `bash -c \"python -c '...'\"` or write a .py file and run it    |\n| `\u002Ftmp\u002F`                           | Git Bash maps `\u002Ftmp\u002F` automatically — scripts work as-is          |\n| `grep -in ... \\| head -20`        | Works inside `bash -c \"...\"`                                      |\n| `brew install --cask libreoffice` | `winget install TheDocumentFoundation.LibreOffice`                |\n| `coreutils` (`gtimeout`)          | Not needed — `recalc.py` falls back to `subprocess` timeout kwarg |\n\n**If Git Bash or any tool is missing**, detect before installing anything:\n\n| Tool        | Detect                                                     | Install (run by the user, not the agent)           |\n| ----------- | ---------------------------------------------------------- | -------------------------------------------------- |\n| Python      | `python --version`                                         | `winget install Python.Python.3.12`                |\n| Git Bash    | `ls \"$PROGRAMFILES\u002FGit\u002Fbin\u002Fbash.exe\"`                      | `winget install Git.Git`                           |\n| LibreOffice | `ls \"$PROGRAMFILES\u002FLibreOffice\u002Fprogram\u002Fsoffice.exe\"`       | `winget install TheDocumentFoundation.LibreOffice` |\n| openpyxl    | `python -c \"import openpyxl; print(openpyxl.__version__)\"` | `python -m pip install openpyxl`                   |\n\nThis fork deliberately does **not** ship an auto-installer. Upstream delegated the bootstrap to a\nreference inside a different skill in its own host product; that skill is not part of ZCodium, so\nthe pointer dangles. A skill that also writes to the user's shell config or installs global\npackages is the wrong default on a managed host anyway. Report what is missing and let the user\nrun the install command.\n",{"name":5,"license":15,"description":7,"descriptions":24},{"zh-Hans":25},"读取、编辑、创建和转换表格文件，支持 xlsx、xlsm、csv、tsv、公式、格式、图表和数据清洗。","main","cf24e347c4aaf1182ef9fa1f9743a80ab14ff638",".agents\u002Fskills\u002Fxlsx",20,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium","https:\u002F\u002Fzcodium-project.github.io\u002F","https:\u002F\u002Fgithub.com\u002FZCodium-project","zcodium-project.github.io\u002F","Zcode Open Audit Version.","2026-11-13 18:44:38",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add ZCodium-project\u002FZCodium --skill xlsx",{"tool":42,"command":43},"bunx","bunx skills add ZCodium-project\u002FZCodium --skill xlsx",{"tool":45,"command":46},"pnpm","pnpm dlx skills add ZCodium-project\u002FZCodium --skill xlsx","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002FSKILL.md",true,[50,56,60,64,68,72,76,80,84,88,94,98,102,106,110,114,118,122,126,130,134,138,142,146,150,154,158,162,166,170,174,178,182,186,190,194,198,202,206,210,214,218,222,226,230,234,238,242,246,250,254,258,262,266,270,274,278,282,286,290,294,298],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"docs\u002Fadvanced-reference.md","REFERENCE","参考文档",19169,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fadvanced-reference.md",{"path":57,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":58,"sourceHtmlUrl":59},"docs\u002Fcases\u002Fsuperstore-multiformat\u002Fconversion_manifest.example.json",1246,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fcases\u002Fsuperstore-multiformat\u002Fconversion_manifest.example.json",{"path":61,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":62,"sourceHtmlUrl":63},"docs\u002Fcases\u002Fsuperstore-multiformat\u002Fsuperstore_orders_print.html",7739991,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fcases\u002Fsuperstore-multiformat\u002Fsuperstore_orders_print.html",{"path":65,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":66,"sourceHtmlUrl":67},"docs\u002Fconventions-guide.md",12870,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fconventions-guide.md",{"path":69,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":70,"sourceHtmlUrl":71},"docs\u002Fcreate-edit-guide.md",15667,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fcreate-edit-guide.md",{"path":73,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":74,"sourceHtmlUrl":75},"docs\u002Fpitfalls-index.md",33351,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fpitfalls-index.md",{"path":77,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":78,"sourceHtmlUrl":79},"docs\u002Fraw-xml-escape-hatch.md",17213,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fraw-xml-escape-hatch.md",{"path":81,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":82,"sourceHtmlUrl":83},"docs\u002Frecalc-guide.md",14803,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Frecalc-guide.md",{"path":85,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":86,"sourceHtmlUrl":87},"docs\u002Fsuperstore-multiformat-conversion-case.md",19711,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fdocs\u002Fsuperstore-multiformat-conversion-case.md",{"path":89,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":92,"sourceHtmlUrl":93},"scripts\u002Foffice\u002Fhelpers\u002F__init__.py","SCRIPT","脚本",0,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fhelpers\u002F__init__.py",{"path":95,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":96,"sourceHtmlUrl":97},"scripts\u002Foffice\u002Fhelpers\u002Fmerge_runs.py",5567,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fhelpers\u002Fmerge_runs.py",{"path":99,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":100,"sourceHtmlUrl":101},"scripts\u002Foffice\u002Fhelpers\u002Fsimplify_redlines.py",5973,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fhelpers\u002Fsimplify_redlines.py",{"path":103,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":104,"sourceHtmlUrl":105},"scripts\u002Foffice\u002Fpack.py",5420,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fpack.py",{"path":107,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":108,"sourceHtmlUrl":109},"scripts\u002Foffice\u002Fredlining_author.py",1798,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fredlining_author.py",{"path":111,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":112,"sourceHtmlUrl":113},"scripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-contentTypes.xsd",1963,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-contentTypes.xsd",{"path":115,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":116,"sourceHtmlUrl":117},"scripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-coreProperties.xsd",2515,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-coreProperties.xsd",{"path":119,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":120,"sourceHtmlUrl":121},"scripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-digSig.xsd",2856,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-digSig.xsd",{"path":123,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":124,"sourceHtmlUrl":125},"scripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-relationships.xsd",1344,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fecma\u002Ffouth-edition\u002Fopc-relationships.xsd",{"path":127,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":128,"sourceHtmlUrl":129},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-chart.xsd",74984,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-chart.xsd",{"path":131,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":132,"sourceHtmlUrl":133},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-chartDrawing.xsd",6956,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-chartDrawing.xsd",{"path":135,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":136,"sourceHtmlUrl":137},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-diagram.xsd",51302,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-diagram.xsd",{"path":139,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":140,"sourceHtmlUrl":141},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-lockedCanvas.xsd",624,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-lockedCanvas.xsd",{"path":143,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":144,"sourceHtmlUrl":145},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-main.xsd",152039,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-main.xsd",{"path":147,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":148,"sourceHtmlUrl":149},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-picture.xsd",1231,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-picture.xsd",{"path":151,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":152,"sourceHtmlUrl":153},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-spreadsheetDrawing.xsd",8862,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-spreadsheetDrawing.xsd",{"path":155,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":156,"sourceHtmlUrl":157},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-wordprocessingDrawing.xsd",14795,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fdml-wordprocessingDrawing.xsd",{"path":159,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":160,"sourceHtmlUrl":161},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fpml.xsd",83612,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fpml.xsd",{"path":163,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":164,"sourceHtmlUrl":165},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-additionalCharacteristics.xsd",1269,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-additionalCharacteristics.xsd",{"path":167,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":168,"sourceHtmlUrl":169},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-bibliography.xsd",7328,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-bibliography.xsd",{"path":171,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":172,"sourceHtmlUrl":173},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-commonSimpleTypes.xsd",6382,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-commonSimpleTypes.xsd",{"path":175,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":176,"sourceHtmlUrl":177},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-customXmlDataProperties.xsd",1248,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-customXmlDataProperties.xsd",{"path":179,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":180,"sourceHtmlUrl":181},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-customXmlSchemaProperties.xsd",880,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-customXmlSchemaProperties.xsd",{"path":183,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":184,"sourceHtmlUrl":185},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesCustom.xsd",2608,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesCustom.xsd",{"path":187,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":188,"sourceHtmlUrl":189},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesExtended.xsd",3507,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesExtended.xsd",{"path":191,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":192,"sourceHtmlUrl":193},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesVariantTypes.xsd",7507,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-documentPropertiesVariantTypes.xsd",{"path":195,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":196,"sourceHtmlUrl":197},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-math.xsd",23313,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-math.xsd",{"path":199,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":200,"sourceHtmlUrl":201},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-relationshipReference.xsd",1367,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fshared-relationshipReference.xsd",{"path":203,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":204,"sourceHtmlUrl":205},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fsml.xsd",242277,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fsml.xsd",{"path":207,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":208,"sourceHtmlUrl":209},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-main.xsd",26148,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-main.xsd",{"path":211,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":212,"sourceHtmlUrl":213},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-officeDrawing.xsd",25279,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-officeDrawing.xsd",{"path":215,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":216,"sourceHtmlUrl":217},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-presentationDrawing.xsd",535,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-presentationDrawing.xsd",{"path":219,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":220,"sourceHtmlUrl":221},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-spreadsheetDrawing.xsd",5712,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-spreadsheetDrawing.xsd",{"path":223,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":224,"sourceHtmlUrl":225},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-wordprocessingDrawing.xsd",4010,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fvml-wordprocessingDrawing.xsd",{"path":227,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":228,"sourceHtmlUrl":229},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fwml.xsd",171367,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fwml.xsd",{"path":231,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":232,"sourceHtmlUrl":233},"scripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fxml.xsd",4646,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002FISO-IEC29500-4_2016\u002Fxml.xsd",{"path":235,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":236,"sourceHtmlUrl":237},"scripts\u002Foffice\u002Fschemas\u002Fmce\u002Fmc.xsd",3127,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmce\u002Fmc.xsd",{"path":239,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":240,"sourceHtmlUrl":241},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2010.xsd",26549,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2010.xsd",{"path":243,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":244,"sourceHtmlUrl":245},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2012.xsd",3745,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2012.xsd",{"path":247,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":248,"sourceHtmlUrl":249},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2018.xsd",901,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-2018.xsd",{"path":251,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":252,"sourceHtmlUrl":253},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-cex-2018.xsd",1778,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-cex-2018.xsd",{"path":255,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":256,"sourceHtmlUrl":257},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-cid-2016.xsd",1002,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-cid-2016.xsd",{"path":259,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":260,"sourceHtmlUrl":261},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-sdtdatahash-2020.xsd",600,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-sdtdatahash-2020.xsd",{"path":263,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":264,"sourceHtmlUrl":265},"scripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-symex-2015.xsd",745,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fschemas\u002Fmicrosoft\u002Fwml-symex-2015.xsd",{"path":267,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":268,"sourceHtmlUrl":269},"scripts\u002Foffice\u002Fsoffice.py",6003,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fsoffice.py",{"path":271,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":272,"sourceHtmlUrl":273},"scripts\u002Foffice\u002Funpack.py",4052,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Funpack.py",{"path":275,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":276,"sourceHtmlUrl":277},"scripts\u002Foffice\u002Fvalidate.py",4275,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidate.py",{"path":279,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":280,"sourceHtmlUrl":281},"scripts\u002Foffice\u002Fvalidators\u002F__init__.py",336,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidators\u002F__init__.py",{"path":283,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":284,"sourceHtmlUrl":285},"scripts\u002Foffice\u002Fvalidators\u002Fbase.py",32859,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidators\u002Fbase.py",{"path":287,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":288,"sourceHtmlUrl":289},"scripts\u002Foffice\u002Fvalidators\u002Fdocx.py",16376,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidators\u002Fdocx.py",{"path":291,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":292,"sourceHtmlUrl":293},"scripts\u002Foffice\u002Fvalidators\u002Fpptx.py",9824,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidators\u002Fpptx.py",{"path":295,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":296,"sourceHtmlUrl":297},"scripts\u002Foffice\u002Fvalidators\u002Fredlining.py",9139,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Foffice\u002Fvalidators\u002Fredlining.py",{"path":299,"resourceType":90,"resourceTypeDesc":91,"sizeBytes":300,"sourceHtmlUrl":301},"scripts\u002Frecalc.py",17429,"https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fxlsx\u002Fscripts\u002Frecalc.py",[303,310,317,324,331,338],{"skillNo":304,"name":305,"nameZh":6,"description":306,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":307,"categories":308,"keywords":309,"publishedAt":19},"01M7DBYM2HAMB3K65Q3XET0XWW","plugin-mcp-ping","E2E workspace Plugin skill marker","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fpackages\u002Fdesktop\u002Ftest\u002Fe2e\u002Ffixtures\u002Fplugin-mcp-skill\u002Fe2e-plugin-mcp-ping\u002Fskills\u002Fplugin-mcp-ping\u002FSKILL.md",[],[],{"skillNo":311,"name":312,"nameZh":6,"description":313,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":314,"categories":315,"keywords":316,"publishedAt":19},"01M7DBYCTV753XKE9TRBT8CRSX","zcode-configuration-guide","Use when configuring ZCode's extension resources (MCP servers, slash commands, skills, hooks, and plugins) or instruction files such as AGENTS.md in the ZCode client. Explains where each resource is configured at the user and workspace scope, the discovery order, precedence, and merge rules, plus guidance on which location to choose. Use when someone asks how to add an MCP server, command, skill, hook, plugin, or AGENTS.md instructions, where a configuration file lives, why a configuration is not taking effect, or needs routing to a specific diagnostic skill.","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fapps\u002Fzcode-cli\u002Fpackages\u002Fzcode-guide-plugin\u002Fskills\u002Fzcode-configuration-guide\u002FSKILL.md",[],[],{"skillNo":318,"name":319,"nameZh":6,"description":320,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":321,"categories":322,"keywords":323,"publishedAt":19},"01M7DBY4P7GEEPQZ5VSWCF0YS1","diagnosing-skills","Use to diagnose and fix ZCode skill configuration problems in the ZCode client. Applies when a skill is not discovered, is installed but does not trigger automatically, is shadowed by a higher-precedence skill of the same name, is disabled by configuration, has a SKILL.md frontmatter error, fails to load because its description is too long, or disappears because the plugin providing it is disabled. Provides the discovery order, how to inspect skills in the client, common pitfalls, and a step-by-step localization and repair workflow.","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fapps\u002Fzcode-cli\u002Fpackages\u002Fzcode-guide-plugin\u002Fskills\u002Fdiagnosing-skills\u002FSKILL.md",[],[],{"skillNo":325,"name":326,"nameZh":6,"description":327,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":328,"categories":329,"keywords":330,"publishedAt":19},"01M7DBXS9PAV8Z09309HMYMNPC","diagnosing-plugins","Use to diagnose and fix ZCode plugin and marketplace problems in the ZCode client. Applies when a plugin is not listed, adding a marketplace or installing a plugin fails, a plugin is enabled but its skills or commands are missing, a built-in plugin still appears after being disabled, a plugin is not enabled as expected, a plugin.json manifest has a parse error, a plugin name is invalid, a dependency is unresolved or blocked across marketplaces, or a sensitive configuration value cannot be entered in the interface. Provides the plugin lifecycle, manifest schema, how to manage plugins in the client, and a step-by-step localization and repair workflow.","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fapps\u002Fzcode-cli\u002Fpackages\u002Fzcode-guide-plugin\u002Fskills\u002Fdiagnosing-plugins\u002FSKILL.md",[],[],{"skillNo":332,"name":333,"nameZh":6,"description":334,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":335,"categories":336,"keywords":337,"publishedAt":19},"01M7DBXHJ0G1KZDC405HYQBHB6","diagnosing-mcp","Use to diagnose and fix ZCode MCP (Model Context Protocol) server configuration problems in the ZCode client. Applies when an MCP server will not connect, its tools (mcp__server__tool) do not appear, it shows as disabled or failed, connections time out, a command cannot be found, template variables are not expanded, or a server defined in a configuration file has no effect. Provides configuration locations, how to inspect status in Settings, common pitfalls, and a step-by-step localization and repair workflow.","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fapps\u002Fzcode-cli\u002Fpackages\u002Fzcode-guide-plugin\u002Fskills\u002Fdiagnosing-mcp\u002FSKILL.md",[],[],{"skillNo":339,"name":340,"nameZh":6,"description":341,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":342,"categories":343,"keywords":344,"publishedAt":19},"01M7DBXABP8ZRNM6TFWDJXJH9Z","diagnosing-hooks","Use to diagnose and fix ZCode hook configuration problems in the ZCode client. Applies when a hook does not trigger, an event name is wrong, a matcher does not match a tool name, a script is not executable, template variables are not expanded, a timeout unit is mistaken (seconds versus milliseconds), the command and process field styles are mixed, a hook's JSON output fails validation, a hook blocks the session unexpectedly, or configuration-file hooks are not enabled. Provides configuration sources, the hooks.json schema, how to inspect hooks in the client, and a step-by-step localization and repair workflow.","https:\u002F\u002Fgithub.com\u002FZCodium-project\u002FZCodium\u002Fblob\u002Fmain\u002Fapps\u002Fzcode-cli\u002Fpackages\u002Fzcode-guide-plugin\u002Fskills\u002Fdiagnosing-hooks\u002FSKILL.md",[],[]]