mcpdata Request an invite

Answers / Get data into your AI

How do I keep a daily snapshot of a CSV export so my AI can compare days?

Recipe verified on Ran on a private test instance
Short answer

Save each day's export with its date in the file name in one folder, and mark that folder as a history folder. cdb-localdata then loads every file as a dated version of a single table, so you can ask for any day, a range of days, or what changed.

Steps

  1. Connect your assistant to your instance (this writes, so it needs your own instance with a ReadWrite key from the accounts portal; the public demo is read-only). In Claude Code: 
    claude mcp add --transport http mcpdata https://[you].mcpdata.org/mcp --header "Authorization: Bearer $CDB_API_KEY"

    Replace [you] with your instance's name. Other apps: Connect your AI.

  2. Install the client and sign it in.
    pip install https://[you].mcpdata.org/help/download/cdb-0.3.4-py3-none-any.whl
cdb-localdata login --to https://[you].mcpdata.org
  3. Mark the exports folder as a history folder. Put a .cdb/config.json in it saying where the date is in each file name (the sample folder exports/daily has it): 
    {"type": "history", "timestamp_from": "filename:%Y-%m-%d"}
  4. Add the folder above it and sync. Each file is sent as a dated version of exports.daily: 
    cdb-localdata add ./samples/exports --to https://[you].mcpdata.org --namespace exports
cdb-localdata sync
  5. Compare two days. In the workbench, or from Python (your AI can save the same SQL as a view for a report): 
    import cdb
s = cdb.Session("https://[you].mcpdata.org")
first = s.execute_sql("SELECT COUNT(*) AS orders FROM exports.daily AS OF LASTDATA BETWEEN #2023-01-01# AND #2023-01-01#")
latest = s.execute_sql("SELECT COUNT(*) AS orders FROM exports.daily AS OF LASTDATA")
print("2023-01-01:", int(first.to_dataframe()["orders"][0]), "orders; latest day:", int(latest.to_dataframe()["orders"][0]), "orders")
  6. Ask your question. Give your AI this prompt:
    Using mcpdata, save a view of orders and revenue by channel on the latest day of exports.daily (AS OF LASTDATA), and make it a daily report.
What your AI does: the tool calls it makes for this prompt, and what came back
create_view(name="exports.latest_by_channel", select_sql="SELECT channel, COUNT(*) AS orders, SUM(revenue) AS revenue FROM exports.daily AS OF LASTDATA GROUP BY channel") → created true, view exports.latest_by_channel
create_report(name="Latest day by channel", view="exports.latest_by_channel") → report_id a760d6d7f5394a07a286bef55c7e5205
run_report(report_id="a760d6d7f5394a07a286bef55c7e5205") → status ok, rows 3
Output of the steps above, from the last run
2023-01-01: 1556 orders; latest day: 1559 orders
Ask Claude with this prompt Ask ChatGPT with this prompt

The assistant needs mcpdata connected (step 1) to run the query.

Why this works

Overwriting yesterday's export loses it; keeping files with different names gives you many tables. A history table keeps one name and one version per day. AS OF LASTDATA BETWEEN #d# AND #d# reads one day, AS OF LASTDATA the latest, and AS OF ALLVERSIONS every day.

Related questions