DataAnthropic
Data
by Anthropic
Cleaning, analysis, and visualization skills for turning a raw dataset into an answer someone can act on.
What’s in the bundle
- Cleaning skills that find the malformed rows before they skew the result
- Exploratory analysis that reports what the data supports
- Visualization guidance matched to the question being asked
- Written interpretation of a result, including its caveats
What you can do with it
- Take a raw export and get a cleaned dataset plus a list of what was wrong with it
- Ask a question of a dataset and get the answer with its confidence stated
- Turn an analysis into a chart and an explanation a stakeholder can read
What it doesn’t do
It works on the extract you provide, not a live warehouse connection. On very large datasets you will be summarizing before it can reason, and any sampling decision is yours to make and state.
Where it runs
Anthropic built Data as a Cowork plugin, which runs inside Claude Desktop. We rebuilt it as MCP-compatible skills, so the same bundle works in Claude, ChatGPT, Cursor, Claude Code, and Agentman over one connection.
Browse all Claude plugins