Move data

Load it once. Query it forever.

mcpdata gives your AI a place to load files and tables of any size once, keep its own working data between conversations, and hand datasets to other agents, so only answers, never raw files, go through the chat.

Files from your computer

cdb-localdata turns folders of CSV, Parquet, ORC and Excel files into tables. Run it once, or let it sync every minute in the background. A history folder keeps every daily file as a dated version of one table, so your AI can compare today with any earlier day. One pip install; your files are only read, and nothing leaves your machine until you sync.

cdb-localdata add "C:\Data\Exports" --to https://[you].mcpdata.org --namespace files
cdb-localdata sync

Loads of any size

Big files are sent in parts automatically. If a part fails it's simply retried, with no duplicate rows. Long loads run as background jobs, so nothing times out half way.

From your code

Upload a table from Python (a pandas DataFrame, a list of rows) with the cdb client, then query it from Python and from your AI alike. Append to a table, or replace it with a new version in one step.

import cdb
s = cdb.Session("https://[you].mcpdata.org")       # the key: ~/.cdb/[you].mcpdata.org.env or CDB_API_KEY
s.write_table("sales.orders", df, replace=True)

Your AI's own tables

Agents write to their own agent.* tables with create_table, append_rows and replace_rows: logs of what they did, snapshots of what they found, state they update. It's all still there in the next conversation, and you can report on it like any other data.

Every version kept

Each load is kept as a version. Ask what a table looked like on any past date, or what changed between two loads.