---
name: datap-find-sellable-data
description: Use when the person asks what data they have that could be sold, or wants to earn from files, exports or logs they already hold. Scans folders the person names for datasets, checks demand on datap and proposes listings.
---

# Find data worth selling on this computer

You need local file access for this. If you cannot read the person's files, say so and ask them to describe or paste what they hold instead.

## 1. Agree the scope first
Ask which folders, drives, databases or cloud exports to look at, and wait for the answer. Never scan a whole home directory or disk unasked. Nothing leaves the computer during this guide: you read names, sizes, dates and column headers, and at most a few rows to understand a file.

## 2. Look for dataset-shaped things
- Tables: CSV, TSV, JSON, NDJSON, Parquet, Excel, SQLite and database dumps.
- Collections: scraped pages, crawl output, API responses saved over time, image, audio or video sets with labels or captions.
- Logs and measurements the person generated: sensor readings, prices they tracked, research results, survey responses they ran, app or game telemetry they own.
- Curated lists: directories, catalogues, taxonomies, benchmark or evaluation sets, prompts with graded answers.
Skip caches, dependencies, system files and anything under a few hundred rows unless it is rare.

## 3. Rights check
Only data the person has the right to sell goes further. Leave out, without opening more than you need to recognise it: personal data about identifiable people (contacts, customers, patients, messages, location history), credentials and keys, an employer's or client's confidential data, and anything bought or downloaded under a licence that forbids redistribution. If ownership is unclear, ask the person; do not guess.

## 4. Score each candidate
For each one note: what a row is, row count, fields, time span, how it was collected, whether it can be refreshed, and who would pay for it. Rate it on:
- Scarcity: can a buyer get this free elsewhere?
- Freshness: is it current, and can it stay current? Maintained data earns more than a one-off file.
- Completeness: stable identifier per row, few gaps, consistent types.
- Demand: call `datatap_list_wishlist` with the topic, and `datatap_search_datasets` to see what is already listed and at what price.

## 5. Report back, briefly
Give a ranked shortlist of at most five: name, one line on what it is, rows, why someone would buy it, the nearest wishlist request or existing listing, and any rights question the person must answer. Say plainly when nothing is worth listing.

## 6. If the person wants to list one
Follow the "List a dataset" guide: `datatap_get_seller_guide` with topic `list-a-dataset`. Upload limits are in `datatap_get_ingest_limits`. Seller overview: https://datap.com/provider

These steps use the datap connector (https://datap.com/mcp/private). Setup: https://datap.com/docs/agents
