**Invoke:** `$codex-skill-creator`

**Notes**
- Creates files only. Writes or edits skill files in the location you choose; never installs, publishes, or deletes skills unless you ask.
- No Tableau connection needed.

Helps you turn a workflow into an installable Codex skill, or improve one you already have. Defines what the skill should and shouldn't handle, writes a precise description so it triggers at the right times, and keeps the package small: a `SKILL.md`, plus references, scripts, or assets only when they add something real. When revising a skill, it targets the actual problem, such as triggering too often or too rarely, conflicting instructions, broken file links, or unsafe write steps. Validates the result with the active Codex skill validator and checks links, commands, and scripts, reporting exactly what was and wasn't tested.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, principles, and expected package layout.
- `agents/openai.yaml`. Codex display name and default prompt.

**What you can ask for**
- "Turn this workflow into an installable skill."
- "Refactor this skill so it triggers less broadly."
- "Move the provider-specific instructions in this skill into reference files."
- "Add a tested helper script for the file transformation this skill describes."

Tableau Skill Creator (Codex)

Creates and maintains focused, validated Codex skills.

Skapat av Tableau + Salesforce

Innehållskvalitet

Lokalisera listningsnamn

Hyper Synthetic Forge generates deterministic synthetic datasets from declarative profiles. It supports related tables, reproducible seeds, CSV/JSON/SQLite output, read-back validation, and manifest-based rerun checks.

## When to use it

* Generate reusable mock data for development, demos, and tests.
* Model multi-table datasets with validated foreign keys.
* Reproduce a dataset exactly from a profile and seed.
* Create output artifacts without copying real records.

Do not use it to query, reproduce, or quality-audit production data.

## Requirements

* Codex with this skill installed or available.
* Python 3.10+.

CSV, JSON, and SQLite generation use only the Python standard library.

## Installation

Install the entire `codex-synth-forge` directory through your normal Codex skill or plugin workflow. Keep `SKILL.md`, `agents/`, and `skill-expertise/` together.

## Quick start

Validate a profile:

```bash
python skill-expertise/synth_forge.py profile-check --profile profile.json
```

Generate a deterministic CSV dataset:

```bash
python skill-expertise/synth_forge.py generate \
  --profile profile.json \
  --output outputs/run-001 \
  --format csv \
  --seed 42
```

Read the generated artifacts back and validate them independently:

```bash
python skill-expertise/synth_forge.py validate \
  --profile profile.json \
  --data outputs/run-001
```

## Example prompt

```text
Use $codex-synth-forge to generate 5,000 customers and 25,000 related orders as CSV. Make email addresses synthetic, preserve referential integrity, and use seed 42.
```

## Output contract

Each successful run includes generated data plus:

* `profile.normalized.json` — canonical input profile;
* `validation.json` — row, type, null, range, uniqueness, relationship, and privacy checks;
* `manifest.json` — run fingerprint and file hashes;
* `GENLOG.md` — compact reproducibility summary.

A run is successful only when `validation.json` reports `ok: true`.

## Privacy model

Profiles may contain aggregate distributions and common non-sensitive categories. They must not contain raw identities, contact details, account identifiers, free text, rare combinations, or source rows. Identity-like values are generated synthetically.

For leakage checks without storing prohibited source values, the profile can contain SHA-256 hashes in `privacy.forbidden_values_sha256`.

## Development

Run the complete test suite and active Codex skill validator before release:

```bash
python -m unittest discover -s skill-expertise -v
```

Hyper Synthetic Forge (Codex)

Generates reproducible synthetic CSV, JSON, and SQLite datasets.

Data och modell

Engelska

Testing Matrix turns a skill, agent, or prompt-driven workflow into realistic behavioral test prompts with explicit routing, pass criteria, and guided result tracking.

## What it tests

* Positive and natural-language activation.
* Nearby requests that should not trigger the target.
* Main-path behavior and promised outputs.
* Missing or ambiguous inputs.
* Dependency failures and safe recovery.
* Authorization, privacy, and destructive-action boundaries.
* Stateful follow-ups, reruns, and regressions.

Tests target observable behavior rather than preferred wording.

## Modes

* **Matrix only:** produce the test plan without running it.
* **Guided run:** walk through one prompt at a time and record results.
* **Results review:** evaluate responses or artifacts supplied by the user.
* **Regression update:** revise coverage after the target changes.
* **Repair and retest:** diagnose failures; target edits still require separate authorization.

For a general “help me test this” request, the skill defaults to a guided standard run.

## Installation

Install the entire `codex-testing-matrix` directory through your normal Codex skill or plugin workflow. Keep `SKILL.md` and `agents/` together.

## Example prompts

```text
Use $codex-testing-matrix to create a standard behavioral test matrix for this skill and guide me through it one prompt at a time.
```

```text
Create release-level regression coverage for this agent, including dependency failures and permission boundaries.
```

```text
Review these target-session responses against the attached test matrix and mark each row Pass, Fail, Blocked, N/A, or Not run.
```

## How guided testing works

The matrix tells the user where each test belongs:

* a fresh target session for independent activation and boundary checks;
* the same target session for deliberate stateful sequences;
* an isolated workspace for scripts and file behavior;
* the target application for product-specific behavior;
* the current conversation for planning and result review.

Expected outcomes remain outside the copy-paste test prompt so the system under test is not coached.

## Result states

Rows are recorded as **Pass**, **Fail**, **Blocked**, **N/A**, or **Not run**. Record fixes and prior failures in the Retest history. Unclear observations stay Blocked with the missing evidence identified. Missing runtime access remains blocked; untested behavior is never converted into a pass.

Tableau Testing Matrix (Codex)

Builds and guides behavioral test matrices for Codex skills and workflows.

**Invoke:** `$pulse-blueprint`

**Notes**

* Changes Tableau content when you ask. Design, audit, and recommendations are read-only; it creates or updates Pulse definitions only when you explicitly ask, and only for the definitions you name.
* Never deletes metrics or changes subscriptions or notifications unless you ask for exactly that.
* Needs Tableau MCP for audits and changes.

Turns business decisions and datasource metadata into a small, actionable Pulse metric program. Designs metric blueprints (measure, time grain, dimensions, filters, insights), recommends and ranks metrics for a datasource, and audits existing definitions with scoring that accounts for anything it couldn't check. When you ask for a change, it checks what the connection can actually do, applies the change, and reads it back to confirm. Can compare an audit against a saved baseline.

**What's in the bundle**

* `SKILL.md`. Instructions the agent loads at runtime.
* `README.md`. Usage guide, examples, and validation.
* `skill-expertise/`. Reference files: `audit-methodology.md`, `audit-scoring-contract.md`, `metric-design.md`, `output-contract.md`, `state-and-comparison.md`, `tableau-pulse-capabilities.md`, `score_audit.py` (audit scoring) and `tests/test_score_audit.py`.

**What you can ask for**

* "Design a Tableau Pulse metric for monthly recurring revenue."
* "Recommend Pulse metrics for this pipeline datasource."
* "Audit my Pulse metric definitions."
* "Optimize our churn metric for weekly operating reviews."
* "Apply the agreed formatting changes to these Pulse definitions."

Tableau Pulse Blueprint (Codex)

Designs, audits, and updates Tableau Pulse metric definitions.

**Invoke:** `$tableau-content-compass`

**Notes**
- Read, advise, and check. Never changes your Tableau content.
- Needs Tableau MCP with content-search or metadata access.

Answers "Do we already have Tableau content for this?" Searches views, workbooks, datasources, projects, and Pulse metrics, then ranks matches from explicit metadata evidence into High, Medium, and Low confidence tiers, with the two strongest reasons for each. Separates strong coverage from partial coverage and real gaps, and treats permission errors as unknown coverage rather than proof that nothing exists. Links only to content Tableau actually returns.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, ranking, and confidence tiers.
- `agents/openai.yaml`. Codex display name and default prompt.
- `references/`. Reference files: `ranking-schema.md`, `search-strategy.md`.
- `scripts/`. `rank_content.py` (relevance ranking) and `tests/test_rank_content.py`.

**What you can ask for**
- "Do we already have Tableau content for customer churn?"
- "Find dashboards that show regional sales trends in the Commercial Analytics project."
- "Find the finance project, then show me its strongest bookings content."

Tableau Content Compass (Codex)

Finds and ranks existing Tableau content that answers a business question.

Analyssupport

**Invoke:** `$tableau-governance-scanner`

**Notes**
- Read, advise, and check by default. Fixes such as renames, description or tag updates, moves, or certification happen only when you explicitly ask, and only for the items you name.
- Never deletes, refreshes, publishes, or changes permissions based on audit findings alone.
- Needs Tableau MCP with read access to the scope you want audited.

Audits a Tableau Cloud or Server site or project for governance health: stale-looking content, adoption signals, certification and lineage, naming, ownership concentration, project organization, performance-risk indicators, and metadata completeness. Every finding has a stable ID, the evidence behind it, and a next step. Scoring is normalized so large sites aren't penalized for size, and it reports any area it couldn't assess. Supports project or site comparisons and saved baselines. It doesn't treat a modified date as data freshness or total views as recent adoption.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, limits, and safety.
- `agents/openai.yaml`. Codex display name, default prompt, and Tableau MCP dependency.
- `references/`. Reference files: `mcp-and-evidence.md`, `methodology.md`, `reporting.md`.
- `scripts/`. `score_governance.py` (scoring) and `test_score_governance.py`.

**What you can ask for**
- "Audit the Finance project with the Standard policy."
- "Run a read-only governance scan of this site and give me an executive summary."
- "Compare these two projects for governance health."
- "Save this scan as our governance baseline."

Tableau Governance Scanner (Codex)

Audits Tableau Cloud or Server governance with evidence-backed findings.

**Invoke:** `$tableau-quest`

**Notes**
- Read, advise, and check. Never changes your Tableau content.
- Tableau MCP optional. Field Quests can use read-only metadata from your site, only with your consent, and never cast real people in negative roles.

A choice-driven simulation that teaches Tableau judgment through realistic technical and organizational consequences, not lectures. Choose a single scenario, a campaign, or a Field Quest grounded in your own site, at lengths from Blitz (3 to 5 decisions) to Deep (30 or more). Scenarios cover performance, governance, metrics, reliability, security, UX, and trust, with no obvious right answers. Ends with a text pixel-art ending and a judgment-based Sparkle evaluation. Quest state stays in the conversation unless you export it.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, safety, and privacy.
- `agents/openai.yaml`. Codex display name, default prompt, and optional Tableau MCP dependency.
- `references/`. Reference files: `scenario-design.md`, `state-and-endings.md`, `tableau-grounding.md`.
- `scripts/`. `validate_state.py` (exported state check) and `test_validate_state.py`.

**What you can ask for**
- "Start a quick Tableau adventure."
- "Start TableauQuest in noir mode with a metric-conflict scenario."
- "Run a Field Quest based on my Tableau project, and anonymize all people."

TableauQuest (Codex)

Choose-your-own-adventure that teaches Tableau judgment through consequences.

**Invoke:** `$tableau-scribe`

**Notes**
- Read, advise, and check by default. Writes descriptions back to Tableau only when you explicitly ask, and only to the fields you name.
- Never queries sample values unless you approve it.
- Needs Tableau MCP.

Creates documentation someone can use without knowing Tableau internals: brief or detailed dashboard help pages, and business-friendly field and metric dictionaries for a datasource. Every claim comes from visible content, Tableau metadata, approved queries, or what you tell it, and anything inferred or unknown is labeled. Can refresh an existing guide against the current dashboard. Delivers Markdown, DOCX, or XLSX, checked against an evidence ledger.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide and examples.
- `agents/openai.yaml`. Codex display name and default prompt.
- `references/`. Reference files: `dashboard-documentation.md`, `datasource-dictionary.md`, `evidence-contract.md`, `output-formats.md`, `tableau-evidence.md`.
- `scripts/`. `validate_evidence.py` (evidence check) and `tests/test_validate_evidence.py`.

**What you can ask for**
- "Document this Tableau dashboard for our sales managers."
- "Create a brief help-page description for this view."
- "Build a metric dictionary for the calculations in this datasource."
- "Create an XLSX data dictionary for all fields."
- "Update this dashboard guide against the current view."

Tableau Scribe (Codex)

Documents Tableau dashboards and builds datasource data dictionaries.

**Invoke:** `$tableau-vibe-code`

**Notes**
- Creates files only by default. Works on a copy and returns a new `.twb` or `.twbx`; never overwrites your original workbook.
- Publishes to Tableau Cloud or Server only when you explicitly ask and name the destination, and protects existing workbooks from being replaced by accident.
- Tableau connection optional. Needed only to open, render, or publish the result.

Builds or restyles Tableau `.twb` and `.twbx` workbooks through careful XML edits. Applies a deliberate restyle (palette, fonts, backgrounds), adds supported worksheets, dashboard zones, or KPI rows using a compatible workbook as the pattern, implements findings from a dashboard critique, or applies the same change across a batch. Safely unpacks packaged workbooks, keeps every datasource and embedded asset, validates the XML after each change, and repackages the result. The changelog separates what was checked locally from what was confirmed in Tableau, and the skill won't invent XML it has no proven pattern for.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, helpers, and safety model.
- `agents/openai.yaml`. Codex display name and default prompt.
- `references/`. Reference file: `workbook-xml-rules.md`.
- `scripts/`. `workbook_tools.py` (extract, validate, recolor, and package) and `tests/test_workbook_tools.py`.

**What you can ask for**
- "Convert this workbook to our dark palette and keep every datasource and image."
- "Apply these dashboard critique findings and return a validated .twbx."
- "Add a KPI row using this other workbook as the pattern. Don't publish anything."
- "Restyle these three workbooks with the same fonts and colors."

Tableau Vibe Code (Codex)

Builds and restyles Tableau workbooks with safe, validated XML edits.

Skapa och designa

**Slash command:** `/hyper-synthetic-forge`

**Notes**
- Creates files. Never changes or publishes your Tableau datasources; publishing a generated `.hyper` is a manual step.
- Tableau MCP optional. Clone mode profiles a published datasource with aggregate queries only and never pulls real rows.

Generates synthetic test data two ways: clone a published Tableau datasource's schema, distributions, correlations, and seasonality, or build from a plain-language description using domain templates (e-commerce, healthcare, financial services, SaaS, supply chain, HR). Shows a schema proposal and a 5-row preview for approval, then generates and validates the data for accuracy, consistency, and privacy. Can inject anomalies (spikes, dips, outliers, outages) or realistic messiness (typos, casing, near-duplicates) with an answer key. Exports to `.hyper` (recommended), CSV, JSON, Excel, or SQL, and can save the profile to regenerate later.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, modes, export formats, and limitations.
- `skill-expertise/`. Reference files: `domain-templates.md`, `error-recovery-playbook.md`, `export-reference.md`, `generation-algorithms.md`, `injection-catalogs.md`, `output-templates.md`, `profiling-recipes.md`, `validation-checklists.md`.

**What you can ask for**
- "Generate test data for an e-commerce store with 10K rows."
- "Clone this datasource with synthetic data."
- "I need fake data with a Black Friday spike to test Pulse alerts."
- "Make a messy version of this data for testing a Prep flow."
- "Forge 50K rows from my saved schema."

Hyper Synthetic Forge (Claude)

Generates realistic synthetic test data as Hyper, CSV, JSON, Excel, or SQL.

**Slash command:** `/pulse-blueprint`

**Notes**
- Read, advise, and check. Never creates, edits, or deletes Pulse metrics; you configure the blueprint in the Pulse UI.
- Needs Tableau MCP.
- Remembers past runs. Saves past audits and blueprints to compare changes and avoid duplicates.

Turns business questions or a datasource into implementation-ready Pulse metric blueprints: measure, time dimension, dimensions, granularities, sentiment, comparisons, filtered metrics, and which insights to turn on, using the exact Pulse settings so you can configure them directly. Four entry points: start from a business question, recommend up to 5 metrics for a datasource, audit existing definitions against 17 checks with a score, or optimize an existing definition against the datasource's fields. Checks for existing definitions first so it doesn't recommend duplicates, and labels an audit Partial when a check can't run.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, audit checks, design principles, and limitations.
- `skill-expertise/`. Reference files: `api-enum-reference.md`, `audit-and-optimize-templates.md`, `blueprint-templates.md`, `memory-file-templates.md`.

**What you can ask for**
- "What should I track for monthly pipeline coverage?"
- "Recommend Pulse metrics for this datasource."
- "Audit my Pulse metrics."
- "Optimize my churn metric."
- "Create a Pulse metric for monthly recurring revenue."

Tableau Pulse Blueprint (Claude)

Designs and audits Tableau Pulse metric definitions as ready-to-build specs.

**Slash command:** `/tableau-calc-engine`

**Notes**
- Read, advise, and check. Gives you paste-ready formulas; you add them in Tableau.
- Tableau MCP optional. When connected, it checks field names against the live schema and tests row-level, aggregate, and LOD logic with read-only queries. Without it, it says validation was skipped and lists test cases to check manually.

Writes, debugs, translates, and optimizes Tableau calculated fields in four modes: Write, Debug, Translate (SQL to Tableau), and Optimize. Every formula comes back commented, correctly typed (row-level, aggregate, LOD, or table calc), placed in Tableau's order of operations, and named in 20 characters or less. Covers LOD expressions, table calcs, sets and set actions, dynamic zone visibility, spatial functions, string and regex functions, parameters, and performance tuning, with ready recipes like running totals, YoY, Top N, and last N complete periods. Flags features that need a newer Tableau version.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, modes, coverage, and limitations.
- `skill-expertise/`. Reference files: `dynamic-zones-and-parameters.md`, `lod-patterns.md`, `naming-and-error-reference.md`, `performance-rules.md`, `recipe-library.md`, `sets-and-set-actions.md`, `spatial-and-string-functions.md`, `table-calc-patterns.md`.

**What you can ask for**
- "Write a calc that shows YoY growth by region."
- "Why is this returning NULL?"
- "Write an LOD expression for first purchase date per customer."
- "Translate this SQL to Tableau."
- "Optimize this calc. It's slow on 2M rows."

Tableau Calc Engine (Claude)

Writes, debugs, translates, and optimizes Tableau calculated fields.

Beräkningsfamilj

**Slash command:** `/tableau-content-compass`

**Notes**
- Read, advise, and check. Never changes your Tableau content.
- Needs Tableau MCP.
- Remembers past runs. Saves searched projects, known gaps, and top results to speed up later searches.

Answers "Do we already have something for this?" Searches a project's views, workbooks, datasource fields, and Pulse metrics, then scores every match from 0 to 100 and groups results into High, Medium, and Low confidence with the exact reasons for each score. Understands related terms (churn, attrition, retention) and infers when a question needs a date field. Helps you find the right project if you don't know it. When nothing matches, says so clearly and suggests next steps. Adds an Open in Tableau link only when Tableau returns one; links are never guessed.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, scoring system, and limitations.
- `skill-expertise/`. Reference files: `api-field-reference.md`, `matching-intelligence.md`, `output-templates.md`, `project-discovery.md`, `scoring-reference.md`.

**What you can ask for**
- "Do we have anything showing customer churn by region?"
- "Which view shows pipeline by quarter?"
- "Find a dashboard for regional sales trends."
- "What projects are there?"

Tableau Content Compass (Claude)

**Slash command:** `/tableau-governance-scanner`

**Notes**
- Read, advise, and check. Never changes content or permissions on your site.
- Creates files. Can save the report as HTML, Excel, Word, Markdown, or JSON.
- Needs Tableau MCP with access to views and workbooks; datasource and project access are optional.
- Remembers past runs. Saves settings per scope so it can show score changes between scans.

Audits a Tableau Cloud or Server site or project across 7 areas: staleness, usage and adoption (including zero-view "ghost" workbooks), certification and trust, naming, structure, performance risk, and metadata. Returns a letter-graded health report with quick wins and longer-term goals. Choose a Light, Standard, or Strict preset, compare two projects or track changes over time, and set up scheduled scans that alert you only when scores drop. If some APIs aren't accessible, it grades those areas N/A and labels the run Partial.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, audit areas, output formats, and limitations.
- `skill-expertise/`. Reference files: `api-field-availability.md`, `calibration-notes.md`, `check-categories.md`, `compare-mode.md`, `output-formats.md`, `scheduled-scans.md`.

**What you can ask for**
- "Audit my site."
- "Run a governance scan on the Finance project."
- "Find ghost workbooks."
- "Compare these two projects."
- "What needs attention on my Tableau site?"

Tableau Governance Scanner (Claude)

Audits a Tableau site or project for governance and hygiene issues.

**Slash command:** `/tableau-insights-service`

**Notes**
- Read, advise, and check. Never creates or edits Pulse metrics or other Tableau content.
- Needs Tableau MCP with Pulse access. Insight calls can take 5 to 25 seconds.

Answers business questions about metric performance, trends, changes, anomalies, drivers, goals, and forecasts using Tableau Pulse's pre-computed insights for governed metrics. Finds the right metric (subscribed metrics rank higher), gathers its full definition, and picks the right insight call for one metric or several. Leads with the direct answer, then two or three supporting signals. When two metrics share a name, it tells you which it used or asks. If no metric fits or the service can't answer, it falls back to querying a datasource and tells you it did and why.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. How it works and when it triggers.
- `skill-expertise/`. Reference files: `insight-payloads.md`, `metric-context-mapping.md`, `question-patterns.md`.

**What you can ask for**
- "Why did revenue change last month?"
- "Are we on track to hit our bookings goal?"
- "How is churn trending?"
- "What changed in my metrics this week?"
- "Summarize my metrics."

Tableau Insights Service (Claude)

Answers metric questions with Tableau Pulse insights before querying raw data.

**Slash command:** `/tableau-quest`

**Notes**
- Read, advise, and check. Never changes your Tableau content.
- Tableau MCP optional. Field Quests can use read-only metadata from your site, or you can paste your real situation instead.

A choice-driven simulation that teaches Tableau judgment through realistic organizational consequences, not tutorials. Drops you into scenarios like a broken dashboard before a board meeting or two teams reporting conflicting numbers, and your choices ripple forward. Play a Single Scenario, a 7 to 10 scenario Campaign where earlier decisions carry over, or a Field Quest grounded in your own site. Pick a genre (from Straightforward Reality to Noir or Office Comedy) and a length from Blitz (3 to 5 decisions) to Deep (30 or more). Every ending includes a pixel art scene and a Sparkle rating for judgment quality, not technical correctness.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, modes, genres, and Sparkle evaluation.
- `skill-expertise/`. Reference files: `blitz-example-turns.md`, `narrative-arc-templates.md`, `player-facing-scripts.md`, `scenario-archetypes.md`, `site-grounded-mode.md`.

**What you can ask for**
- "Tableau quest."
- "Teach me Tableau through a scenario."
- "Start a Blitz adventure in noir mode."
- "Run a Field Quest based on my Tableau site."

TableauQuest (Claude)

**Slash command:** `/worksheet-cleanup`

**Notes**
- **Changes your workbook, with your consent.** Removes leftover worksheets only after you approve each one, and writes to a copy of the workbook by default.

Finds leftover worksheet scaffolding (hidden test sheets, scratch calcs, dead field placements, orphan filters) and removes it only with your consent for each item, writing to a copy by default. Always runs a read-only inventory first, so you see everything before anything is removed.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `rule_catalog.md`.

**What you can ask for**
- "Clean up worksheets."
- "Remove orphan sheets."
- "Audit unused worksheets."
- "Prune this workbook."
- "Remove hidden test sheets I don't need."

Worksheet Cleanup

Removes leftover scratch and orphan worksheets, with your consent.

Publicera och driftsätt

**Slash command:** `/workbook-optimizer`

**Notes**
- ***Runs a Python script on your machine*** (Python 3.9 or later). Needs a .twb or .twbx saved on disk.
- Changes a copy, never your original. Fixes are applied only after you say yes, and each one is saved as a new numbered copy.

Scans a Tableau workbook for load-time slowdowns against a catalog of 50+ known issues, such as automatic dashboard sizing, too many filter cards, unused fields and redundant calcs. Groups findings into Warnings, Notes and Info, each with a cited rule and a recommended action, plus an Optimization Readiness Score and a shareable HTML report. Safe mechanical fixes, like setting a fixed dashboard size or hiding unused fields, can be applied one at a time or all at once. Every fix is validated, and nothing is written if a change would break the workbook.

**What’s in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, modes, example and limitations.
- `skill-expertise/`. The `wbo.py` tool and its rule catalog.

**What you can ask for**
- "Optimize this workbook."
- "Why is this workbook slow?"
- "Show me stats on this workbook."
- "Check for redundant calcs."

Workbook Optimizer

Audits workbook performance and applies approved fixes to a copy.

**Slash command:** `/workbook-auditor`

**Notes**
- Read, advise, and check. Never changes your workbook. Saves a snapshot file so it can compare later.

Snapshots a workbook's fields, calcs, sheets, and dashboards, then compares the current state to a saved baseline and reports what changed in plain English. Useful for governance and collaborator handoffs. Writes only the snapshot file.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `report_template.md`, `snapshot_schema.md`.

**What you can ask for**
- "What changed in this workbook?"
- "Did anyone edit this calc?"
- "Compare to the last version."
- "Snapshot this workbook."
- "Establish a baseline so I can track changes."

Workbook Auditor

Tracks what changed in a workbook since your last snapshot.

**Slash command:** `/sheet-swap-detector`

**Notes**
- Read, advise, and check. Never changes your workbook.

Scans every worksheet in a workbook, groups near-duplicates into 5 cluster types (filter, aggregation, sort, chart-type variants, and old versions), and recommends how to consolidate each cluster with numbered Desktop steps. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `cluster-taxonomy.md`, `consolidation-patterns.md`.

**What you can ask for**
- "Find duplicate worksheets."
- "Which sheets can I consolidate?"
- "My worksheet list is a mess."
- "Find redundant sheets."

Sheet Swap Detector

Finds near-duplicate worksheets and shows how to consolidate them.

**Slash command:** `/publish-readiness`

**Notes**
- Read, advise, and check. Never changes your workbook.

Runs a pre-publish sweep before a workbook goes to Server or Cloud. Separates findings the agent verified (unused fields, default names, missing titles, scratch sheets, performance risks, calc hygiene) from a checklist only a person can confirm. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `agent_limitations.md`, `performance_checks.md`, `publishing_checklist.md`.

**What you can ask for**
- "Am I ready to publish?"
- "Pre-publish check."
- "Clean this up before I publish."
- "Anything to fix before this goes live?"

Publish Readiness

Runs a pre-publish checklist before a workbook goes live.

**Slash command:** `/dashboard-tab-order`

**Notes**
- Read, advise, and check. Never changes your workbook.

Audits the order of a workbook's dashboard tabs for a logical left-to-right flow against 8 navigation rules, including entry-point clarity, drill-down grouping, and orphan tabs. Read-only, with drag-to-fix steps.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `rule_catalog.md`.

**What you can ask for**
- "Audit my dashboard tab order."
- "Check tab flow."
- "Is my navigation logical?"
- "Tab strip audit."
- "Does my tab order make sense for a first-time viewer?"

Dashboard Tab Order

Checks that dashboard tabs follow a logical order.

**Slash command:** `/dashboard-print-check`

**Notes**
- Read, advise, and check. Never changes your workbook.

Checks a dashboard for print-readiness: aspect ratio, zone density per page, black-and-white color safety, and text legibility at print scale. Read-only. Every finding is a manual Desktop fix step.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `check_catalog.md`.

**What you can ask for**
- "Is this dashboard print-ready?"
- "Print check."
- "Audit this dashboard for print."
- "Will this look OK on Letter paper?"
- "Will this print well in black and white?"

Dashboard Print Check

Checks whether a dashboard is ready to print.

**Slash command:** `/dashboard-preflight`

**Notes**
- Read, advise, and check. Never changes your workbook.

Audits one dashboard's assembly before it ships: zone alignment, sheet mapping, filter actions, parameter linkage, titles and captions, canvas size, and view fit. Read-only. Every finding is a Desktop fix instruction.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `handoff_schema.md`, `preflight_catalog.md`, `report_template.md`, `validation_gate.md`.

**What you can ask for**
- "Preflight this dashboard."
- "Check my dashboard before I publish."
- "Is this dashboard ready for review?"
- "Dashboard sanity check."
- "Audit this dashboard's assembly."

Dashboard Preflight

Checks a dashboard's assembly before it ships.

**Slash command:** `/join-and-relationship-advisor`

**Notes**
- Read, advise, and check. Never changes your workbook.

Recommends a join or a relationship for two tables, checks grain and cardinality, and flags measures that would double or rows that would explode before they corrupt a dashboard. Advisory only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `fanning_patterns.md`, `join_vs_relationship.md`, `output_template.md`.

**What you can ask for**
- "Join or relationship?"
- "Why is my measure doubling?"
- "Check cardinality."
- "Grain check on Orders and Returns."
- "Why is SUM(Sales) doubling?"

Join and Relationship Advisor

Recommends joins vs. relationships and catches doubling measures.

**Slash command:** `/geocoding-auditor`

**Notes**
- Read, advise, and check. Never changes your workbook.

Audits a workbook's geographic fields for wrong geographic roles, high rates of unmatched or ambiguous locations, and missing hierarchies that silently break maps. Read-only. Every fix is a Desktop right-click step.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `geocoding_checks.md`, `output_template.md`, `severity_rubric.md`.

**What you can ask for**
- "Why is my map wrong?"
- "Unknown locations on my map."
- "Why won't my cities match?"
- "Check my geographic roles."
- "Why are marks missing from my map?"

Geocoding Auditor

Finds geographic role and location problems that break maps.

**Slash command:** `/extract-optimizer`

**Notes**
- Read, advise, and check. Never changes your workbook.

Reviews an extract's configuration for refresh speed and storage across four areas (incremental refresh key, aggregation, hidden-field pruning, extract filters) and recommends changes with their expected impact. Advisory only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `extract_dimensions.md`, `impact_rubric.md`, `output_template.md`.

**What you can ask for**
- "Why is my extract so slow?"
- "Why is my extract so big?"
- "Optimize this extract."
- "Set up incremental refresh."
- "Should I aggregate this extract?"

Extract Optimizer

Recommends extract settings for faster refreshes and smaller files.

**Slash command:** `/esri-map-builder`

**Notes**
- Runs Python scripts on your machine. If the `tableauhyperapi` package is missing, it asks once before installing it, and stops if you decline.
- Needs the Esri JDBC driver. If it’s missing, the skill gives you the download link; it never downloads the driver for you.
- Public ArcGIS services only. Services behind sign-in aren’t supported.
- Creates a new workbook; nothing existing is changed.

Turns an ArcGIS REST service (MapServer or FeatureServer) into a Tableau workbook, with one map layer per service layer stacked on a single sheet. Give it a service URL, or a folder or ArcGIS Online org plus a theme like “voting” or “flood”. It groups the available layers so you can pick by meaning, combines several services in one workbook, and turns tables without geometry into separate data sources. Layers start as one-row placeholders. Run **Extract > Refresh All Extracts** and save to load the data.

**What’s in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, requirements and known issues.
- `skill-expertise/`. Reference notes and the scripts `check_driver.py`, `list_layers.py` and `gen_workbook.py`.

**What you can ask for**
- "Map this ArcGIS service: [MapServer URL]."
- "Show me the flood layers in this ArcGIS Online org and build a map."
- "Recreate this ArcGIS map in Tableau."

Esri Map Builder

Builds a Tableau map workbook from ArcGIS service layers.

**Slash command:** `/data-source-connector`

**Notes**
- Read, advise, and check. Never changes your workbook. Gives you Desktop steps to make the connection yourself.

Recommends the right Tableau Desktop connector and sign-in method (OAuth, PAT, password, key pair) for a new data source, with prerequisites and numbered setup steps. Advisory only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `auth_patterns.md`, `connector_catalog.md`, `output_template.md`.

**What you can ask for**
- "How do I connect to Snowflake?"
- "Which connector for Postgres?"
- "OAuth, PAT, or password?"
- "Help me set up a new data source."
- "How do I connect to Postgres with SSO?"

Data Source Connector

Recommends the right connector and sign-in method for new data.

**Slash command:** `/data-scribe`

**Notes**
- **Temporary changes only.** Adds a scratch sheet to sample values, then deletes it. Saves the metric dictionary as an .md or .txt file.

Generates a field-level metric dictionary from a connected data source: what each field is, how each calc works, sample values, and which sheets use each field. Writes one Markdown or text file. One data source per run.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `output_template.md`.

**What you can ask for**
- "Document this data source."
- "Build a metric dictionary."
- "What do these fields mean?"
- "Create a data dictionary."
- "Export field documentation as a Markdown file."

Data Scribe

Builds a metric dictionary explaining every field in a data source.

**Slash command:** `/data-quality-review`

**Notes**
- **Changes your workbook only with your approval.** Proposes reversible fix calcs and adds them only when you say yes. Uses a temporary scratch sheet to read the data, then removes it.

Profiles a connected data source for quality issues in two layers: a metadata pre-scan that runs in seconds (naming, type and role mismatches, near-duplicates, unused fields) and a row-level scan of the data (nulls, duplicates, whitespace, outliers, dates). Every proposed fix is reversible and needs your approval.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `checks.md`, `fixes.md`, `metadata_checks.md`.

**What you can ask for**
- "Check data quality."
- "Any nulls or dupes in this data?"
- "Profile this data source."
- "Find data issues."
- "Check my fields for naming problems."

Data Quality Review

Profiles a data source for nulls, duplicates, outliers, and naming issues.

**Slash command:** `/data-model-documenter`

**Notes**
- Read, advise, and check. Never changes your workbook. Saves the data model as a markdown file.

Exports a data source's logical model (tables, relationships, join conditions, and calc dependencies) to a shareable markdown file. Read-only plus one file write. A current-state snapshot, with no comparison to earlier versions.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `output_template.md`.

**What you can ask for**
- "Document this data model."
- "Export the tables and relationships."
- "Share this data model."
- "Make a data model doc for handoff."
- "Document the data source structure."

Data Model Documenter

Exports a data source's tables and relationships to a shareable file.

**Slash command:** `/testing-matrix`

**Notes**
- Read, advise, and check. Never changes your workbook, which is used read-only as a test fixture. Saves test files, and edits the skill being tested only with your approval.

Builds a test plan for a Tableau Desktop agent skill: 20-25 prompts across 5-8 sequences covering its triggers, rules, and error handling. Walks you through each in Desktop, and on a failure proposes a fix and applies it only with your approval.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `fix_patterns.md`, `sequence_library.md`.

**What you can ask for**
- "Test this skill."
- "Build a testing matrix."
- "QA plan for this skill."
- "Regression-test this skill after my last edit."

Testing Matrix

Builds and runs a test plan for a Tableau Desktop agent skill.

**Slash command:** `/desktop-spellcheck`

**Notes**
- Read, advise, and check. Never changes your workbook.

Reads every piece of stakeholder-visible text in a workbook (titles, captions, dashboard text, aliases, calc and parameter names, tooltips, annotations), checks it against a dictionary and your allow-list, and produces a severity-scored report. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `allow-list.txt`. User-maintained allow-list of accepted terms.
- `skill-expertise/`. Reference files: `example_clean_report.md`, `report_template.md`, `severity_rubric.md`, `test_prompts.md`, `xml_extraction_patterns.md`.

**What you can ask for**
- "Spellcheck this workbook."
- "Find typos."
- "Proofread this dashboard."
- "Check my copy."

Spellcheck

Proofreads all user-visible text in a workbook.

**Slash command:** `/interaction-validator`

**Notes**
- **Temporary changes only.** Briefly changes filter and parameter states to test the wiring, then puts them back. Creates no worksheets, calcs, or actions.

Checks dashboard wiring after a build: filter and set actions, parameters, dynamic zone visibility, and cross-data-source filters. Reports which interactions respond, which are dead, and which need a manual check. Creates no worksheets, calcs, or actions.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `report_format.md`, `scope.md`.

**What you can ask for**
- "Test my actions."
- "Is this filter working?"
- "Nothing happens when I click."
- "Does this parameter do anything?"
- "Verify my dashboard wiring."

Interaction Validator

Tests dashboard actions, parameters, and filters for broken wiring.

**Slash command:** `/i18n-scanner`

**Notes**
- Read, advise, and check. Never changes your workbook. Saves a translation-ready CSV and a findings report.

Finds localization risks before a workbook ships to a global audience, such as hard-coded currency symbols, US-style dates, and English text inside calcs. Produces a translation-ready CSV and a severity-ranked report. Read-only. Your translator can work from the CSV without opening Desktop.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.

**What you can ask for**
- "Is this dashboard ready for global?"
- "Find hard-coded text."
- "Export text for translation."
- "Localize this dashboard."
- "Is this workbook ready for our EU customers?"

i18n Scanner

Finds localization risks and exports text for translation.

**Slash command:** `/format-drift-scanner`

**Notes**
- Read, advise, and check. Never changes your workbook.

Audits a workbook for inconsistent number, date, currency, and font formats across measures, date fields, and text objects. Read-only. Every finding is a Desktop fix instruction. Catches mismatched currency symbols, decimal places, thousands separators, and negative-number styles.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `drift_catalog.md`, `handoff_schema.md`, `report_template.md`, `validation_gate.md`.

**What you can ask for**
- "Check my formatting."
- "Are my dates and numbers consistent?"
- "Format drift scan."
- "Check number formats."
- "Date format audit."

Format Drift Scanner

Finds inconsistent number, date, currency, and font formats.

**Slash command:** `/dashboard-caption-audit`

**Notes**
- Read, advise, and check. Never changes your workbook.

Audits a workbook for missing or default titles and captions across dashboards, worksheets, text objects, and phone or tablet layouts. Findings are grouped by severity with Desktop fix steps. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `completeness_catalog.md`, `handoff_schema.md`, `report_template.md`.

**What you can ask for**
- "Check my titles."
- "Audit captions."
- "Are my dashboards labeled?"
- "Find missing titles in this workbook."
- "Which sheets still have default names?"

Dashboard Caption Audit

Finds missing or default titles and captions across a workbook.

**Slash command:** `/chart-swap-advisor`

**Notes**
- Read, advise, and check. Never changes your workbook.

Evaluates one worksheet's chart type against the question it answers and recommends a better chart when the fit is poor, with the reason. Blocks known anti-patterns such as 3D charts and pies with more than 5 slices. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `anti_patterns.md`, `chart_recommendation_catalog.md`, `handoff_schema.md`, `question_type_taxonomy.md`, `report_template.md`, `validation_gate.md`.

**What you can ask for**
- "Is this the right chart type?"
- "Should this be a different chart?"
- "Recommend a better chart."
- "Evaluate this chart."

Chart Swap Advisor

Checks whether a chart fits its question and suggests a better one.

**Slash command:** `/caption-and-tooltip-advisor`

**Notes**
- Read, advise, and check. Never changes your workbook. Proposes rewrites for you to apply in Desktop.

Reads an open dashboard and rewrites weak tooltips, captions, and labels for your audience. Each finding shows the current text, the proposed replacement, and the numbered Desktop step to apply it. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `audience-voice-rules.md`, `tooltip-catalog.md`.

**What you can ask for**
- "Rewrite my tooltips."
- "Fix my captions."
- "My tooltips look weird."
- "SUM of Sales in the tooltip looks ugly."
- "Rewrite the captions for an executive audience."

Caption and Tooltip Advisor

Rewrites weak tooltips, captions, and labels for your audience.

**Slash command:** `/calc-debugger`

**Notes**
- **Changes your workbook.** Rewrites the broken calc with the fix. May add a temporary scratch sheet to check the result, then removes it.

Diagnoses one broken Tableau calculated field per run (NULLs, wrong values, syntax errors, silent failures) and proposes a corrected version. Every fix passes an 8-check Validation Gate and is verified against your data before delivery.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `error_patterns.md`, `handoff_schema.md`, `validation_gate.md`, plus a function library (`INDEX.md` and files for aggregate, date, LOD, logical, number, spatial, string, table calc, and type conversion).

**What you can ask for**
- "Debug this calc, it returns NULL for some customers."
- "Why is my profit margin calc showing negative values?"
- "Fix this broken calc."
- "My LOD returns the wrong grain, help me fix it."
- "This calc had a syntax error after I edited it."

Calc Debugger

Finds why a calculated field is broken and fixes it.

**Slash command:** `/calc-documenter`

**Notes**
- Read, advise, and check. Never changes your workbook.

Generates structural documentation for every calculated field in a workbook: formula, fields referenced, what uses it, the sheets and dashboards that depend on it, and its aggregation level. One run produces one dependency-sorted dictionary, ready to hand off or archive. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `aggregation_patterns.md`, `documentation_template.md`, `handoff_schema.md`.

**What you can ask for**
- "Document my calcs."
- "Generate a calc dictionary."
- "Map my calc dependencies."
- "Which calcs do I have?"
- "Show me which calcs feed other calcs."

Calc Documenter

Documents every calculated field in a workbook.

**Slash command:** `/calc-explainer`

**Notes**
- Read, advise, and check. Never changes your workbook.

Explains a single Tableau calculated field in plain language: what it computes, why it's built that way, how it fits with the workbook's other calcs, and how it behaves at runtime. Grounded in the actual formula. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `explanation_patterns.md`, `handoff_schema.md`.

**What you can ask for**
- "Explain this calc."
- "What does Profit Ratio do?"
- "Walk me through this calculated field."
- "Why is this calc built this way?"
- "Explain this LOD in plain English."

Calc Explainer

Explains any calculated field in plain language.

**Slash command:** `/calc-hygiene-scan`

**Notes**
- Read, advise, and check. Never changes your workbook. Gives you Desktop steps to fix findings yourself.

Audits every calculated field in a workbook against an 8-rule hygiene catalog: naming drift, missing captions, unused calcs, deep nesting, hard-coded values, deprecated functions, unexplained cross-datasource calcs, and LOD overuse. Read-only. Every finding comes with manual Desktop fix steps.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `rule_catalog.md` and `explanation_patterns.md`.

**What you can ask for**
- "Audit my calcs."
- "Calc hygiene scan."
- "Check calc quality."
- "Find calc issues."
- "Find unused calcs and hard-coded values."

Calc Hygiene Scan

Audits all calcs for naming, unused fields, and other hygiene issues.

**Slash command:** `/calc-inventory`

**Notes**
- Read, advise, and check. Never changes your workbook.

Catalogs every calculated field in one data source per run: name, type, formula, references, style and performance flags, and where each calc is used. Delivers a scannable table, with batched pagination for data sources with 20+ calcs. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `catalog_schema.md`, `handoff_schema.md`, `validation_gate.md`, plus a 10-file `functions/` library covering aggregate, date, LOD, logical, number, spatial, string, table calc, and type conversion.

**What you can ask for**
- "List all calcs in this data source."
- "Inventory my calcs."
- "What calcs exist here?"
- "Catalog my calculated fields."
- "Show me every calc in this workbook."

Calc Inventory

Catalogs every calculated field in a data source.

**Slash command:** `/calc-linter`

**Notes**
- **Changes your workbook only if you opt in.** By default it only reports style issues. If you ask for a rewrite, it replaces the calc after checking it returns the same values.

Lints one existing calculated field per run against a 12-rule house-style catalog (naming, comments, indentation, clarity) and reports each violation with its severity and reason. Rewrites only when you opt in, and every rewrite is checked to return the same values.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `handoff_schema.md`, `style_catalog.md`, `validation_gate.md`, plus a 10-file `functions/` library covering aggregate, date, LOD, logical, number, spatial, string, table calc, and type conversion.

**What you can ask for**
- "Lint this calc."
- "Check this calc's style."
- "Rewrite this calc for readability."
- "Does this calc follow our house style?"
- "Check my calc naming."

Calc Linter

Checks a calc against house style and rewrites it on request.

**Slash command:** `/calc-migrator-diff-report`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Produces a per-calc before-and-after report from a Calc Migrator run: source and target formulas, field remaps, validation status, and anything that went stale after the migration. Read-only. Use it right after a migration to see which calcs need follow-up.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `diff_catalog.md`.

**What you can ask for**
- "Diff my migration."
- "Review migration output."
- "Show migration changes."
- "What did the migration change?"
- "Audit my migration."

Calc Migrator Diff Report

Shows exactly what changed in each calc after a migration.

**Slash command:** `/calc-migrator`

**Notes**
- **Changes your workbook, with your consent.** Adds a copy of the calc to the target data source. The original calc is never changed.

Migrates one calculated field from a source data source to a target data source in the same open workbook. Remaps the fields it references, asks for consent before writing, and verifies the result. The source calc is never changed.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `aggregation_preservation.md`.

**What you can ask for**
- "Migrate this calc."
- "Move this calc to another data source."
- "Port this calc to another data source."
- "Copy my profit calcs to the new data source."

Calc Migrator

Copies a calculated field to another data source in the same workbook.

**Slash command:** `/calc-optimizer`

**Notes**
- **Changes your workbook.** Replaces the calc with the faster version after confirming it returns the same values. Uses a temporary scratch sheet for that check, then removes it.

Optimizes one existing calculated field per run for performance without changing what it returns. Ranks the calc against a 5-tier performance hierarchy, proposes a rewrite or confirms it's already optimal, and verifies the rewrite returns identical values before delivery.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `handoff_schema.md`, `performance_hierarchy.md`, `validation_gate.md`, plus a 10-file `functions/` library covering aggregate, date, LOD, logical, number, spatial, string, table calc, and type conversion.

**What you can ask for**
- "Optimize this calc."
- "This calc is slow."
- "Make this calc faster."
- "Speed up this calculation."
- "Why is this calc slow?"

Calc Optimizer

Makes a slow calculated field faster without changing its results.

**Slash command:** `/calc-translator`

**Notes**
- **Changes your workbook.** Adds the translated calc and checks it against your data, using a temporary scratch sheet that it removes afterward.

Translates a single expression from SQL, Excel, or Power BI DAX into a working Tableau calculated field. Picks the right calc type, writes the calc, verifies it against real data, and says plainly where the translation isn't exact.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.

**What you can ask for**
- "Translate this Excel formula."
- "Convert this DAX to Tableau."
- "Translate this SQL."
- "What would this look like in Tableau?"
- "Move this Power BI measure to Tableau."

Calc Translator

Converts SQL, Excel, or DAX expressions into Tableau calcs.

**Slash command:** `/calc-writer`

**Notes**
- **Changes your workbook.** Adds the new calc. For higher-risk calcs it also adds a temporary scratch sheet to check the result, then removes it.

Authors one new Tableau calculated field per run from a plain-language ask. Every calc passes an 8-check Validation Gate before delivery, and higher-risk calcs (LODs, table calcs, date logic) are also checked against your data. Each reply says whether the calc was verified.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `domain_accuracy.md`, `handoff_schema.md`, `scale_and_modes.md`, `validation_gate.md`, plus a 10-file `functions/` library covering aggregate, date, LOD, logical, number, spatial, string, table calc, and type conversion.

**What you can ask for**
- "I need a calc that computes profit margin."
- "Write a calc for revenue by customer."
- "Create a calc that flags orders over $500."
- "Make a calc for days between order and ship date."
- "Create a rolling 90-day average sales calc."

Calc Writer

Writes a new Tableau calculated field from plain language.

**Slash command:** `/worksheet-builder`

**Notes**
- **Changes your workbook.** Builds the worksheet, and may add the calcs it needs.

Turns a plain-language question into the best-fit chart. Reads your data source, picks the chart type, checks it against 15 chart anti-patterns, and either builds the worksheet or gives click-by-click steps.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `anti_patterns.md`, `build_patterns.md`, `calc_templates.md`, `color_accessibility.md`, `decision_framework.md`, `dual_axis_charts.md`, `filter_context_gaps.md`, `kpi_highlight_box.md`, `number_formatting.md`, `parameters_known_gaps.md`, `worked_examples.md`.

**What you can ask for**
- "Build a chart."
- "Show me sales by region."
- "What chart should I use?"
- "Make a worksheet."
- "Show me the top 10 customers by revenue."

Worksheet Builder

Turns a plain-language question into the best-fit chart.

**Slash command:** `/viz-planner`

**Notes**
- **Changes your workbook.** Builds the worksheets and calcs for the dashboard. Dashboard assembly itself is given as Desktop steps for you to follow.

Plans a complete dashboard from a data source, business questions, and an audience: chart choices, layout, filters, actions, color, and KPIs. Coordinates building the sheets and calcs, then gives exact Desktop steps to assemble the dashboard.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `anti_patterns.md`, `build_patterns.md`, `calc_templates.md`, `color_accessibility.md`, `container_patterns.md`, `decision_framework.md`, `device_layouts.md`, `filter_action_wiring.md`, `filter_interaction.md`, `kpi_highlight_box.md`, `kpi_patterns.md`, `layout_patterns.md`, `worked_examples.md`.

**What you can ask for**
- "Plan a dashboard."
- "Design a dashboard for my VP."
- "What should I build?"
- "Wireframe a KPI, regional breakdown, and trend view."

Viz Planner

Designs a complete dashboard from a business question and audience.

**Slash command:** `/viz-critique-pro-studio`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Scores a dashboard across 7 areas (audience, message, chart choice, layout, color, accessibility, interactivity) plus an ethics check for misleading visuals. Attach a screenshot for full scoring. Gives a short chat summary and an optional full report. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `audience_and_context.md`, `chart_selection_guide.md`, `color_typography_accessibility.md`, `ethics_privacy_gatekeeper.md`, `handoff_schema.md`, `interactive_elements.md`, `layout_and_hierarchy.md`, `report_template.md`, `rubric_evaluation_framework.md`, `scoring_tiers_and_tone.md`, `validation_gate.md`.

**What you can ask for**
- "Critique this dashboard."
- "Score this dashboard."
- "How does this dashboard rate?"
- "Review this viz."

VizCritique Pro

Scores a dashboard across 7 areas and ranks the fixes by impact.

**Slash command:** `/phone-layout-blueprint`

**Notes**
- Read, advise, and check. Never changes your workbook. Gives you Desktop steps to apply yourself.

Designs or audits a dashboard's phone layout. Blueprint mode plans one from your desktop dashboard; audit mode checks an existing layout against 15 rules. Every finding comes with numbered Desktop steps. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `desktop-ui-fix-paths.md`, `phone-rule-catalog.md`, `preference-flag-steps.md`.

**What you can ask for**
- "Make this dashboard mobile-friendly."
- "Audit my phone layout."
- "Why does my Tableau Mobile look broken?"
- "Turn off phone layout auto-generation."

Phone Layout Blueprint

Plans or audits a dashboard's phone layout with step-by-step fixes.

**Slash command:** `/workbook-summary-onepager`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Produces a one-page summary of a workbook for someone who won't open Desktop: what it shows, key metrics, data sources, dashboards, and known limitations. Read-only, with an optional file save. Uses other skills' documentation when it's available in the session.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference file: `section_catalog.md`.

**What you can ask for**
- "Summarize this workbook."
- "One-pager please."
- "Stakeholder summary."
- "Hand off this workbook."
- "Write a one-pager for a colleague who won't open Desktop."

Workbook Summary Onepager

A one-page workbook summary for people who won't open Desktop.

**Slash command:** `/what-if-advisor`

**Notes**
- Read, advise, and check. Never changes your workbook. Gives you Desktop steps to apply yourself.

Recommends the right built-in way to show a hypothetical, such as a reference line, band, trend extension, set-based scenario, or scenario calc, and gives numbered Desktop steps. Avoids parameter authoring, which is weak in Studio today. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `mechanism-catalog.md`, `question-classification.md`.

**What you can ask for**
- "What if we doubled marketing spend?"
- "Add a scenario."
- "Project this sales trend out 6 months."
- "Add a $1M target line to this chart."
- "Run a what-if."

What-If Advisor

Recommends the simplest way to show a what-if scenario in Tableau.

**Slash command:** `/stakeholder-translator`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Turns a dashboard into a one-page plain-language summary for a non-technical audience: what it answers, what it doesn't, three likely follow-up questions, and the chart that carries the headline. Written for the audience you choose. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `jargon-swap-list.md`, `stakeholder-voice-guide.md`.

**What you can ask for**
- "Translate this for my VP."
- "Give me a plain-language version."
- "Explain this to a non-analyst."
- "Jargon-free summary."

Stakeholder Translator

Turns a dashboard into a plain-language summary for non-technical readers.

**Slash command:** `/logic-explainer`

**Notes**
- Read, advise, and check. Never changes your workbook.

Explains how a worksheet or dashboard is built: not just which filters, calcs, LODs, parameters, sets, and actions are on it, but why each one is there and what it does for that view. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `construct_pattern_library.md`, `output_template.md`, `worked_examples.md`.

**What you can ask for**
- "Explain this dashboard's logic."
- "Why does this work this way?"
- "Walk me through this worksheet."
- "What is this dashboard doing?"

Logic Explainer

Explains how and why a worksheet or dashboard is built the way it is.

**Slash command:** `/inheritance-triage`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Orients you in a workbook you just inherited. Orient mode gives a short summary: key dashboards, data sources, calc complexity, and where to start reading. Complexity mode ranks every worksheet from 1 to 10 against a 6-factor rubric. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `complexity-rubric.md`, `triage-output-template.md`.

**What you can ask for**
- "I inherited this workbook."
- "Where do I start?"
- "Orient me."
- "Which sheets are hardest?"
- "Give me a 5-minute summary."

Inheritance Triage

Get oriented fast in a workbook you just inherited.

**Slash command:** `/calc-orchestrator`

**Notes**
- Read, advise, and check. Never changes your workbook.

Routes an unclear calculation request to the right calc skill: writing, debugging, optimizing, linting, or inventory. Asks one clarifying question when two could fit, then hands off with the exact skill to use. Has no calc logic of its own.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `handoff_schema.md`, `peer_routing.md`, `validation_gate.md`.

**What you can ask for**
- "Help me with a calc."
- "I have a calc question."
- "Which skill handles this calc?"
- "I need to do something with this calc, but I'm not sure what."

Calc Orchestrator

Routes unclear calculation requests to the right calc skill.

**Slash command:** `/dashboard-walkthrough-generator`

**Notes**
- Read, advise, and check. Never changes your workbook. Can save its results as an .md or .txt file.

Reads an open dashboard and writes a presenter script: walkthrough order, speaker notes for each sheet, a 30-second elevator version, and the questions your audience is most likely to ask. Read-only.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, examples, limitations, troubleshooting.
- `skill-expertise/`. Reference files: `audience-voice-guide.md`, `script-templates.md`.

**What you can ask for**
- "Write a walkthrough script."
- "Give me presenter notes."
- "How do I present this?"
- "Talking points for this dashboard."
- "What questions will the exec team ask?"

Dashboard Walkthrough Generator

Writes a presenter script and speaker notes for any dashboard.

**Slash command:** `/tableau-scribe`

**Notes**
- Read, advise, and check. Never changes your Tableau content.
- Creates files. Delivers on screen, or as a Word (`.docx`) or Markdown file.
- Needs Tableau MCP with at least Viewer access to the content you document.

Generates documentation from live Tableau content, with no screenshots or copy-paste. Two modes: Auto-Doc documents a published view (purpose, key metrics, filters, how to use it, data and refresh, a short design health summary, and a glossary), and Metric Builder turns a datasource into a plain-language metric dictionary with definitions, direction, and field types. Choose Brief (one page) or Detailed. If a non-essential call fails, it still delivers and explains the gaps in a closing note, marking anything it couldn't confirm. If it can't reach the view itself, it stops and says so.

**What's in the bundle**
- `SKILL.md`. Instructions the agent loads at runtime.
- `README.md`. Usage guide, output structure, and error handling.
- `skill-expertise/`. Reference files: `auto-doc-structures.md`, `design-health-summary.md`, `edge-cases.md`, `error-handling-notes.md`, `identification-resolution-patterns.md`, `metric-dictionary-structures.md`, `output-format-handling.md`.

**What you can ask for**
- "Document the Sales Overview dashboard as a Word doc."
- "Generate a metric dictionary for this datasource."
- "Give me a brief reference card for this view."
- "What does this dashboard show?"

Tableau Scribe (Claude)

Documents Tableau dashboards and builds metric dictionaries from live content.

Codex Calc Engine helps Codex create, debug, translate, and optimize Tableau calculated fields. It combines field discovery, grain analysis, safe authoring, stored-formula read-back, and behavioral checks so a successful API response is not mistaken for a correct calculation.

## When to use it

* Turn a business definition into a Tableau calculated field.
* Repair mixed-aggregation, LOD, null, or table-calculation problems.
* Translate supported SQL expressions into Tableau semantics.
* Review an expensive calculation without changing its meaning.
* Author a dependency-ordered batch of calculations.

Use a workbook-authoring skill for dashboard layout, visual styling, or broad `.twb`/`.twbx` changes.

## Requirements

* Codex with this skill installed or available.
* A Tableau Desktop agent or MCP runtime for live discovery, authoring, and behavioral verification.
* Python 3.10+ for the optional local formula checker.

Without Tableau access, the skill can still draft formulas, but it will label storage and behavior as unverified.

## Installation

Install the entire `codex-calc-engine` directory through your normal Codex skill or plugin workflow. Keep `SKILL.md`, `agents/`, and `skill-expertise/` together.

## Example prompts

```text
Use $codex-calc-engine to create a 90-day retention rate using [User ID], [Order Date], and [First Order Date].
```

```text
Debug this Tableau calculation. It returns “cannot mix aggregate and non-aggregate”: SUM([Sales]) / [Customer Target].
```

```text
Translate this SQL window calculation into Tableau and explain the required table-calculation addressing.
```

## Local formula checks

Normalize a candidate formula for comparison:

```bash
python skill-expertise/calc_checks.py normalize --file formula.txt
```

Check a stored formula for common transport problems:

```bash
python skill-expertise/calc_checks.py check --file stored-formula.txt --expect-operator '<'
```

The checker is intentionally side-effect free. Tableau remains the authority for formula validity and workbook behavior.

## Verification model

The skill distinguishes three outcomes:

1. **Drafted** — a formula was produced without live authoring.
2. **Stored** — Tableau's stored definition was read back and matched.
3. **Behaviorally verified** — representative workbook results were checked at the intended grain.

Blocked checks remain blocked; the skill does not infer success from missing runtime access.

## Development

Run the helper tests and active Codex skill validator before release:

```bash
python -m unittest discover -s skill-expertise -v
```

Tableau Calc Engine (Codex)

Authors and verifies Tableau calculations with safe formula read-back.

Connect Tableau to Anaplan and unlock enterprise planning analytics. Access data from Anaplan's cloud-based connected planning platform directly in Tableau for finance, supply chain, workforce, and sales planning analysis.

The Anaplan connector enables extract-based connectivity to Anaplan models, workspaces, and data sources. Analyze planning scenarios, forecast accuracy, budget vs. actuals, and workforce allocation with the full power of Tableau visualizations while your data remains governed and secure within Anaplan.

Leverage SQL-92 query capabilities to filter, aggregate, and join Anaplan planning data with other enterprise sources for cross-functional analytics and reporting.

Typ av erbjudande

Funktioner

Kategorier

Sökresultat