Add excel-wps-table-diagnosis skill

skills/excel-wps-table-diagnosis/README.md

@@ -0,0 +1,42 @@
+# Excel / WPS Table Diagnosis
+
+Inspect table structure first, then suggest practical Excel/WPS formulas and cleanup workflows.
+
+## What It Does
+
+This skill is designed for common spreadsheet tasks that usually start with messy data rather than a clean formula problem.
+
+It helps with:
+
+- table diagnosis from headers and sample rows
+- cleanup planning for blanks, duplicates, and mixed formats
+- lookup and matching workflows
+- summarization guidance
+- practical Excel and WPS compatibility choices
+- common formula troubleshooting
+
+## Working Style
+
+The skill is meant to:
+
+1. inspect the table first
+2. explain what should be cleaned or checked
+3. recommend a practical workflow
+4. pause before direct spreadsheet work when user approval is needed
+5. continue with concrete formulas or steps after approval
+
+## Files
+
+- `SKILL.md`: entry file
+- `references/workflow.md`: longer working notes
+- `references/use-cases.md`: common task shapes
+- `references/approval-flow.md`: example of diagnose first, execute after approval
+- `references/table-diagnosis.md`: messy table example
+- `references/sheet-match.md`: matching example
+- `references/text-cleanup.md`: text cleanup example
+
+## Typical Prompts
+
+- `Inspect this table and tell me what should be cleaned first.`
+- `Suggest a WPS-friendly lookup formula for these two sheets.`
+- `Check this formula workflow and tell me whether helper columns would be safer.`

skills/excel-wps-table-diagnosis/SKILL.md

@@ -0,0 +1,85 @@
+---
+name: excel-wps-table-diagnosis
+description: Diagnose spreadsheet structure and recommend practical Excel/WPS-compatible formulas and workflows for cleaning, lookup and matching, summarization, text and date cleanup, and common formula troubleshooting. Use when the user asks what formula to use, how to process a table, how to clean messy columns, how to match sheets, how to make a solution work in both Excel and WPS, or when headers and sample rows should be inspected before deciding what to do next.
+---
+
+# Excel/WPS Table Diagnosis
+
+Use this skill when a spreadsheet task should start with the table itself, not with a guess at the formula.
+
+## Core Approach
+
+1. Inspect the input.
+
+- Identify whether the user provided headers, sample rows, CSV content, or only a task description.
+- If the structure is unclear, ask for the smallest missing detail that would change the recommendation.
+
+2. Diagnose the table.
+
+- Identify likely column types such as IDs, dates, amounts, names, status fields, phone numbers, or free text.
+- Look for blanks, duplicates, mixed formats, unstable lookup keys, and columns that should be cleaned before formulas are applied.
+
+3. Choose the most practical path.
+
+- Prefer simple formulas when they are stable and readable.
+- Prefer Excel and WPS compatible approaches over newer functions with weaker compatibility.
+- Prefer helper columns when a single long formula would be fragile.
+- Recommend built-in spreadsheet tools when they are a better fit than formulas.
+
+4. Present a clear recommendation.
+
+- State what the table appears to contain.
+- State what should be done first.
+- Recommend the formula or workflow.
+- Mention compatibility notes for Excel and WPS.
+- Provide a fallback when the preferred formula may not work everywhere.
+
+5. Pause before execution.
+
+- If the next step would directly modify a workbook, add helper columns, rewrite formulas, or otherwise move from diagnosis into execution, ask for confirmation first.
+- Once the user agrees, continue with the concrete formulas, helper columns, or execution steps instead of repeating the analysis.
+- If the scope is already clear, finish the approved execution step before suggesting extra optional follow-up work.
+
+## Priorities
+
+- Diagnose before suggesting formulas.
+- Prefer maintainability over cleverness.
+- Treat compatibility as an early constraint.
+- Do not force everything into one formula.
+- Do not make direct spreadsheet changes without user confirmation.
+
+## Output Shape
+
+Keep the response close to this shape:
+
+### Diagnosis
+
+- what the table likely contains
+- what looks inconsistent or risky
+
+### Recommendation
+
+- what to do first
+- which formula or method to use
+
+### Execution
+
+- what can be done immediately after approval
+- which formulas, helper columns, or spreadsheet steps should be applied
+
+### Compatibility
+
+- whether it should work in Excel
+- whether it should work in WPS
+- what fallback to use if needed
+
+### Notes
+
+- where to place the formula
+- whether helper columns or built-in tools would be easier
+
+## References
+
+- Read `WORKFLOW.md` for the longer working notes.
+- Read `docs/use-cases.md` for common task shapes.
+- Read `examples/README.md` for example inputs and outputs.

skills/excel-wps-table-diagnosis/references/approval-flow.md

@@ -0,0 +1,66 @@
+# Example: Approval Flow
+
+## Scenario
+
+The user provides a small customer table and asks what should be cleaned first.
+
+Headers:
+
+- Customer ID
+- Name
+- Phone
+- Status
+
+Sample rows:
+
+- 001, Alice, 138 0013 8000, active
+- 002, Bob, 13800138001, Active
+- 002, Bob, 13800138001, active
+- 003, Carol, , inactive
+
+## Phase 1: Diagnose And Recommend
+
+Expected shape:
+
+- explain that `Customer ID` should stay as text
+- point out mixed phone formatting
+- point out inconsistent `Status` casing
+- flag the duplicate `Customer ID`
+- suggest helper columns and a cleanup order
+- pause before execution
+
+Example pause:
+
+`I have not modified anything yet. If you want, I can continue with the exact helper-column setup and the spreadsheet steps.`
+
+## Phase 2: Execute After Approval
+
+Once the user agrees, the response should move straight into execution.
+
+Expected shape:
+
+- specify which helper columns to add
+- give the exact formulas
+- explain where to place them
+- explain fill-down and cleanup steps
+- avoid repeating the same diagnosis unless needed for clarity
+
+Example execution elements:
+
+- `E1: Clean Phone`
+- `F1: Clean Status`
+- `G1: Dedupe Key`
+- `H1: Duplicate?`
+
+- `E2: =SUBSTITUTE(SUBSTITUTE(TRIM(C2),CHAR(160),\"\"),\" \",\"\")`
+- `F2: =LOWER(TRIM(D2))`
+- `G2: =A2&\"|\"&B2&\"|\"&E2`
+- `H2: =IF(COUNTIF($G:$G,G2)>1,\"Duplicate\",\"Keep\")`
+
+## Purpose
+
+This example exists to show the intended behavior of the repository:
+
+1. diagnose first
+2. pause before direct spreadsheet work
+3. continue immediately once the user approves

skills/excel-wps-table-diagnosis/references/sheet-match.md

@@ -0,0 +1,38 @@
+# Example: Sheet Match
+
+## Input
+
+Sheet A:
+
+- Customer ID
+- Name
+- Region
+
+Sheet B:
+
+- Customer ID
+- Last Purchase Date
+- Total Amount
+
+User request:
+
+`I need to match Sheet A and Sheet B by customer ID in WPS.`
+
+## Expected Response Shape
+
+### Diagnosis
+
+- `Customer ID` is the most likely key column
+- matching should be done only after checking for duplicate IDs
+
+### Recommendation
+
+- use a lookup formula based on `Customer ID`
+- if broad compatibility matters, prefer `INDEX + MATCH`
+- if the environment supports it, a simpler lookup function may also be suggested
+
+### Compatibility
+
+- provide a WPS-friendly formula first
+- mention a newer Excel-only alternative only as a secondary option
+

skills/excel-wps-table-diagnosis/references/table-diagnosis.md

@@ -0,0 +1,44 @@
+# Example: Table Diagnosis
+
+## Input
+
+Headers:
+
+- Order ID
+- Customer ID
+- Order Date
+- Amount
+- Status
+
+Sample rows:
+
+- 1001, C001, 2026/03/01, 89.5, Paid
+- 1002, C001, 03-02-2026, 120, paid
+- 1003, , 2026.03.05, 45, Pending
+- 1003, C004, 2026/03/05, 45, Pending
+
+User request:
+
+`Please inspect this table and tell me what I should clean first.`
+
+## Expected Response Shape
+
+### Diagnosis
+
+- `Order Date` uses mixed date formats
+- `Customer ID` has blank values
+- `Order ID` contains duplicates
+- `Status` is inconsistent because `Paid` and `paid` should likely be standardized
+
+### Recommendation
+
+- standardize `Status` first
+- fix blank `Customer ID` values if possible
+- normalize `Order Date` into one format
+- investigate duplicate `Order ID` before summarizing data
+
+### Compatibility
+
+- prefer simple cleanup formulas and helper columns
+- avoid depending on newer dynamic-array functions for this workflow
+

skills/excel-wps-table-diagnosis/references/text-cleanup.md

@@ -0,0 +1,32 @@
+# Example: Text Cleanup
+
+## Input
+
+Column values:
+
+- ` Alice`
+- `Bob  `
+- `  Carol`
+- `DAVE`
+- `eve `
+
+User request:
+
+`This column has extra spaces and inconsistent formatting. How should I clean it?`
+
+## Expected Response Shape
+
+### Diagnosis
+
+- extra leading and trailing spaces
+- mixed capitalization
+
+### Recommendation
+
+- trim extra spaces first
+- then normalize capitalization if needed
+- use helper columns if the cleanup should stay easy to review
+
+### Compatibility
+
+- prefer basic text functions that work in both Excel and WPS

skills/excel-wps-table-diagnosis/references/use-cases.md

@@ -0,0 +1,60 @@
+# Use Cases
+
+## 1. Cleanup Before Analysis
+
+Use the project to inspect a table before writing formulas.
+
+This is usually the least exciting step and also the one that prevents the most avoidable trouble.
+
+Typical signals:
+
+- blank values in key columns
+- inconsistent date formats
+- duplicate identifiers
+- status fields with inconsistent capitalization
+
+## 2. Matching Across Sheets
+
+Use the project when two sheets need to be matched by a shared field such as:
+
+- customer ID
+- employee ID
+- order number
+- product code
+
+The goal is to choose a reliable lookup workflow that still works in Excel and WPS.
+
+If the key column is messy, that problem should be handled first instead of hoping the lookup will be forgiving.
+
+## 3. Text Cleanup
+
+Use the project to normalize columns that contain:
+
+- extra spaces
+- hidden separators
+- mixed capitalization
+- text fragments that should be split or extracted
+
+This is where helper columns usually age better than one heroic formula.
+
+## 4. Summarization
+
+Use the project when the user needs to:
+
+- count records by status or department
+- summarize sales or amounts
+- classify rows by conditions
+- prepare a table for a pivot workflow
+
+The useful question here is often not just how to summarize the table, but whether the table is even ready to be summarized.
+
+## 5. Formula Troubleshooting
+
+Use the project when the spreadsheet already has formulas, but they return:
+
+- missing matches
+- obvious reference mistakes
+- unstable behavior after copy-down
+- errors caused by dirty source columns
+
+Sometimes the formula is wrong. Sometimes the table is wrong. It helps to separate those two problems early.