MCP Data Analysis #241
Description
Feature Request: MCP Data Analysis Tool (MVP)
Is your feature request related to a problem or unsupported use case?
The assistant currently cannot analyze user-uploaded tabular files (CSV / Excel) deterministically nor generate charts based on those files.
There is no data_analysis tool available in the project today, which blocks common use cases such as:
- exploring CSV / Excel files,
- asking natural language questions on data,
- visualizing results with charts.
Describe the solution you'd like
Add a new MCP-based data_analysis tool, backed by a dedicated MCP server, able to:
-
Load CSV and Excel files (including multi-sheet files),
-
Interpret natural language questions,
-
Run deterministic analysis using
pandas/numpy, -
Generate charts when relevant (
matplotlib/seaborn, -
Return:
- an analysis text,
- optionally a generated plot.
An MVP is already in progress:
-
MCP server (work in progress):
https://github.com/etalab-ia/conversations-mcp -
App-side integration (feature branch):
https://github.com/suitenumerique/conversations/tree/camand/feat_mcp_data_analysis
Describe alternatives you've considered
-
Not supporting tabular data analysis for now
→ Limits assistant capabilities for common data-related use cases. -
Implementing ad-hoc analysis directly in the app
→ Harder to maintain and scale.
Using an MCP server provides a cleaner separation and a safer execution environment.
Discovery, Documentation, Adoption, Migration Strategy
User experience
- User uploads a CSV / Excel file
- User asks a question about the data
- Assistant responds with an analysis and, if relevant, an embedded chart
No UI change required for the first version.
Adoption
- Can be enabled progressively (feature flag)
- No migration required for existing conversations
Documentation can be added once the MVP is validated.
Do you want to work on it through a Pull Request?
Yes ✅
I’m already working on the implementation and will coordinate before going further.
Examples
