Data Q&A Assistant

Problem

Personal data is often split across chats, receipts, spreadsheets, and databases. Manual cleanup creates duplicates and missed entries, making questions like “how much did I spend on food this month?” hard to answer. This skill targets lightweight personal analytics, turning entry, deduplication, tagging, and querying into a natural-language workflow.

How It Works

The core loop is parse data → check uniqueness → decide how to handle → persist and confirm. It accepts text, voice, photo OCR, Excel, CSV, and external database inputs. Local storage uses SQLite under ~/.workbuddy/data/data-qa/, with schema decisions made automatically and uniqueness enforced by primary keys plus a time + business tag + core value check. Built-in tags cover dining, transport, sales, procurement, and related categories, with AI-generated subtags for finer slices. Query features include single-record lookup, filtering, aggregation, month-over-month and year-over-year comparisons, extremes, and trend or comparison charts. It fits expense logs, small-business metrics, and personal ledgers.

Boundaries

This is a local personal-data tool, not an enterprise BI or warehouse system. Complex permissions, large-scale concurrency, and cross-database federation still need external infrastructure. Sensitive data relies on local masking, encrypted local storage, and temporary cache cleanup, so regular backups of the data directory are advisable. Suspected duplicates prompt confirmation to prevent accidental insertion or deletion.