Clean spreadsheets
inside your assistant.

The MCP server that gives LLMs specialized primitives to audit, format, and fix messy data directly in Claude, ChatGPT, and Cursor.

$ scan_dataset(file="users.csv")
[info] found 42 whitespace issues
[info] found 12 invalid emails
[info] found 3 missing phone formats
$ apply_cleaning_plan(auto=true)
[success] transformation complete. 248ms.
891 rows / 2s
2,051 rows / 12s
89 tests green
https://dc-mcp-26zy.onrender.com/mcp

Terse transformations.
Clean results.

Agentic Data Cleaner identifies structural anomalies and normalizes them into analysis-ready formats without you writing a single line of Pandas code.

safe_auto_fixes

Standard normalization for dates, currencies, and emails.

approval_flow

Risky deletions or fuzzy matching require human sign-off.

diff --git users.csv
@@ -10,3 +10,3 @@
  id,name,email,spent
- 104, " John Doe ", john.d@gmial.com, NULL
+ 104, "John Doe", john.d@gmail.com, N/A
  105, "Sarah P.", sarah@corp.com, 1200.00
- 106, "mike ", , $45.0
+ 106, "Mike", N/A, 45.00
clean_dataset
One-shot cleaning for small, common datasets.
scan_dataset
Heuristic investigation for quality issues.
get_cleaning_plan
Generate actionable human-readable plan.
apply_cleaning_plan
Execute all items in a validated plan.
apply_cleaning_action
Execute a single atomic transformation.
preview_dataset
Fetch sample rows to inspect current state.
export_dataset
Serialize working state to file storage.
create_upload_url
Secure temp endpoint for local ingestion.
delete_dataset
Memory management for server-side environments.
list_cleaning_actions
Query library of available fix patterns.
tier_free
$0
  • + 5 files / day
  • + 2MB CSV limit
  • + full tool access
self_host
managed
tier_pro
Free/mo
  • + 25MB file limit
  • + all formats (xlsx/parquet)
  • + audit reports
  • + replay scripts
start_managed

Ready to connect.

$ https://dc-mcp-26zy.onrender.com/mcp

or connect to managed endpoint: https://dc-mcp-26zy.onrender.com/mcp