table-analysis-pro

A reusable AgentSkill for professional tabular-data analysis, safe multi-table workflows, and table-analysis project review.

Why this exists

Many table-analysis tasks fail not because of syntax, but because of weak workflow: wrong joins, unclear metric definitions, missing grain checks, and overconfident conclusions.

table-analysis-pro is designed to make those failures less likely by enforcing a more disciplined analysis process.

What it does

table-analysis-pro captures a disciplined workflow for:

• analyzing CSV / Excel / DataFrame-style tabular data
• handling single-table and multi-table tasks safely
• reviewing table-analysis agents, data copilots, and spreadsheet-oriented AI tools
• extracting reusable patterns from table-analysis systems

It focuses on workflow, judgment, and output quality rather than teaching pandas syntax.

Typical use cases

Use this skill when the task involves:

• analyzing CSV / Excel / DataFrame-style data
• checking joins before merging tables
• defining KPIs before computing results
• reviewing a table-analysis agent, workflow, or project
• presenting findings with evidence, caveats, and next steps

Example prompts

• Analyze this CSV and summarize the main KPIs.
• Compare these two tables and identify anomalies.
• Review this table-analysis agent design for join-safety and metric-definition risks.
• Help me validate whether this metric is defined correctly before I compute it.

Quick start

Get the packaged skill

Download the latest .skill artifact from the GitHub Release page:

• v0.1.0 release
• Direct download: table-analysis-pro-v0.1.0.skill

Use the repository directly

If you want to inspect or adapt the skill before packaging, start with:

• SKILL.md for the core workflow
• references/ for reusable analysis and review guidance
• PUBLISHING.md for release and packaging notes

Repository structure

• SKILL.md — core skill instructions
• references/workflow-patterns.md — reusable analysis and review patterns
• references/project-review-rubric.md — checklist for evaluating table-analysis systems
• references/output-patterns.md — response and presentation templates
• references/join-safety.md — safe merge/join guidance
• references/metric-definition.md — KPI and derived-metric guidance
• references/related-projects.md — external comparison points and inspiration
• PUBLISHING.md — release and packaging notes
• CHANGELOG.md — release history

Related project reference

• fishsure/TabClaw — a strong external reference point for conversational table-analysis UX, planning, multi-table orchestration, memory, and skill-learning patterns

Use this skill when the task involves

• schema inspection
• filtering / lookup
• aggregation / KPI summaries
• derived metrics
• pivoting / crosstabs
• merging or comparing tables
• anomaly or data-quality checks
• reviewing a table-analysis workflow or project

Working principles

• inspect before concluding
• clarify when ambiguity matters
• plan before multi-step execution
• verify join keys and grain before merging
• define metrics before computing or comparing them
• prefer structured operations before custom code
• present results with evidence and caveats

Packaging

A distributable .skill package is included in dist/ for release builds.

Release status

v0.1.0 — initial publishable draft.