Data analysis has always been a loop: write a query, check the result, tweak the query, visualize the output, clean the data, repeat. Agent skills collapse that loop. Instead of switching between a SQL client, a notebook, a charting tool, and a data quality script, you hand each step to a dedicated skill running inside your coding agent.

The skills below cover the full data workflow — from query generation through cleaning, visualization, and schema management — built for agents like Claude Code, Codex, and Cursor.

What to Look For

Dialect awareness. SQL is not one language. A skill that generates valid Postgres but breaks on BigQuery or Snowflake will cost you debugging time. The best query skills understand dialect differences and let you specify your target.

Data format handling. Real-world data arrives as CSV, JSON, API responses, and database dumps. Skills that only work with one format force pre-processing before analysis even starts. Look for skills that handle the formats you actually use.

Output quality. A chart skill that produces a generic bar graph for every dataset is not useful. Look for skills that choose the right visualization for the data shape and handle axis formatting, color scales, and responsive layout without manual intervention.

Safety defaults. Database skills that generate destructive operations without warning are a liability. The best skills enforce guardrails: no dropping columns without deprecation, no locking migrations on large tables, rollback scripts generated automatically.

Top Agent Skills for Data Analysis and SQL

1. SQL Generator

The most direct skill in this category. SQL Generator takes a plain-language question about your data and returns a correct, optimized query. It supports PostgreSQL, MySQL, SQLite, BigQuery, and Snowflake dialects out of the box.

What sets it apart is optimization awareness. It does not just write a query that works — it writes one that performs. CTEs, window functions, proper join strategies, and index-aware filtering all come standard. Ask it to “show monthly revenue by product, ranked” and you get a query with partitioned window functions and clean aggregation, not a nested subquery mess.

Compatible with: Claude Code, Codex, Cursor, Universal Category: Data Install: gh skill install alirezarezvani/claude-skills/sql-generator

2. Data Cleaner

Raw data is never clean. Data Cleaner identifies and fixes the problems that derail analysis before you start: missing values, duplicates, inconsistent formatting, outliers, and schema mismatches. It works with CSV, JSON, and SQL tables.

The skill scans a dataset and returns a report of what it found — not just “there are duplicates” but which records, how many, and what the resolution options are. Date normalization, currency formatting, categorical standardization — all handled. Run it before any analysis and you skip the hour of manual wrangling that usually opens every data project.

Compatible with: Claude Code, Codex, Cursor, Universal Category: Data Install: gh skill install ComposioHQ/awesome-claude-skills/data-cleaner

3. Chart Generator

Analysis without visualization is just numbers in a terminal. Chart Generator produces publication-quality charts and dashboards from data, outputting Vega-Lite specs, D3 snippets, Chart.js configs, or Recharts components from plain-language descriptions or raw data.

The skill picks the right chart type for the data shape. Time-series revenue data becomes a line chart with proper axis formatting. Categorical comparisons become grouped bar charts with clear labels. It handles responsive layout and color scales automatically, producing output ready to drop into a dashboard or report.

Compatible with: Claude Code, Cursor, Windsurf Category: Data Install: gh skill install vega/vega-lite

4. Amplitude Analytics

For teams using Amplitude, this official Anthropic skill removes the context switch between your agent and your analytics dashboard. Amplitude Analytics lets you query event data, funnels, and retention metrics directly from your coding agent.

The practical value is speed. Instead of opening Amplitude, building a funnel, waiting for results, then copying numbers into a document, you describe what you want in plain language. “Show me the onboarding funnel drop-off for the last 30 days” returns the data inline, without leaving your terminal.

Compatible with: Claude Code, Cursor Category: Data Install: gh skill install anthropics/knowledge-work-plugins/amplitude

5. Schema Documenter

Database schemas grow undocumented. New columns get added, old ones get repurposed, and the original intent gets lost. Schema Documenter generates human-readable documentation from database schemas, TypeScript types, Pydantic models, and OpenAPI specs.

Point it at a Postgres schema and it returns a data dictionary with column descriptions, relationships, and constraints. Point it at TypeScript interfaces and it documents the type hierarchy for onboarding. The output is structured enough for a team wiki but readable enough for a new hire on day one.

Compatible with: Claude Code, Codex, Cursor Category: Data Install: gh skill install ComposioHQ/awesome-claude-skills/schema-documenter

6. API Response Parser

Most data analysis starts with data from somewhere else: an API, a webhook, a third-party service. API Response Parser handles the messy work of normalizing API responses. It parses nested JSON, flattens inconsistent field names, handles pagination envelopes, and outputs normalized schemas ready for downstream use.

If you regularly pull data from multiple vendors, this skill saves real cleanup time. The output is a clean, flat structure you can feed directly into SQL Generator or Chart Generator without manual transformation.

Compatible with: Claude Code, Codex, Cursor, Universal Category: Data Install: gh skill install sickn33/antigravity-awesome-skills/api-response-parser

How to Choose

Start with where you spend the most time. If writing queries is the bottleneck, SQL Generator alone will change your workflow. If you spend more time cleaning data than analyzing it, Data Cleaner is the first install.

For full-stack data workflows, the combination of SQL Generator + Data Cleaner + Chart Generator covers the core loop: query, clean, visualize. Add Schema Documenter if your team is growing and documentation debt is accumulating.

Amplitude Analytics is specific to Amplitude users but nearly eliminates the tool-switching overhead for product analytics teams. API Response Parser is the right pick if your data sources are external APIs rather than internal databases.

All six skills work inside your existing agent. No new tools, no new dashboards. The data stays in your workflow from start to finish.

FAQ

Q: Can these skills connect to my production database? A: The skills generate queries and process data — they do not manage database connections directly. Your agent environment handles the connection. SQL Generator outputs the query; you run it against whatever database client your project already uses.

Q: Do I need all six skills for a data workflow? A: No. Start with one or two that match your bottleneck. SQL Generator and Data Cleaner cover the most common needs. Add others as your workflow expands.

Q: How do these compare to using a Jupyter notebook? A: Notebooks are great for exploratory work with visual output at each step. Agent skills are better when you want analysis embedded in a larger workflow — generating a report, cleaning data before ingestion, or building a dashboard component. They complement notebooks rather than replacing them.