DataDrop
Sell

Quality checks and going live

The checks a listing goes through in review, the verified badge, and what ends a review.

Draft: not yet checked against the product

When you press Submit for review, the listing moves to Review and DataDrop starts checking it. The listing is locked while the checks run. They run in the background, one at a time: the worker looks for listings to check once a minute.

The checks

The checks come in three groups. Which ones run depends on what the listing holds: the listing's own checks always run, and an asset's checks run once for each asset of that kind in the listing.

Required to publish

A listing goes live only when every one of these has passed, or DataDrop has waived it.

CheckRuns onPasses when
Listing metadatathe listingits title, one-liner and description are all filled in
Secrets scanthe listingno credential or key is found in the listing's text
Personal data scanthe listingno personal data is found in the listing's text
Repository snapshoteach codebasea complete copy of the repository is stored
Git historyeach codebasethe copy carries its commit history: send the repository with its .git folder
Pull requests capturedeach codebasethe copy came from a connected code host, with its pull requests
Tasks presenteach taskthe task is linked from Ember Data at a fixed version
Clean QA resultseach taskthe task has QA results in Ember Data, and none of its runs failed

The two scans read every text of the listing that a buyer reads: its title, one-liner and description, and each asset's title, description, tags and languages. A finding names the kind of thing found and the field it is in, never the value itself.

Pull requests are captured only from a connected code host. A codebase uploaded as a ZIP fails Pull requests captured, and the listing then goes live only once DataDrop waives that check.

Verified badge

These are recommended. A listing can go live without them, but it earns the Verified badge only when every one of them has passed.

CheckRuns onPricePasses when
Description matches the datathe listing2 creditsan AI review finds no claim in the listing's text that the figures DataDrop measured contradict
Description tells a buyer what they getthe listing1 creditan AI review finds that the description says what the data is, where it comes from and what it suits
Tests in the repositoryeach codebasefreethe repository has test files of its own

The AI reviews read only the listing's text and the figures DataDrop measured, never a file of the data. A priced check is charged once, before it runs. Its credits are given back when it ends On our side, when it runs out of attempts, or when it is skipped because the listing was submitted again, left review, or had its text changed after it was scanned; if they cannot be given back at once, its result says the refund is owed. A review that ran but whose answer could not be read ends as Couldn't finish and keeps its charge. It is skipped, with nothing charged, when the organization's credits do not cover its price. Once a required check has failed, every recommended check is skipped.

Insights

Graded from A to D, with no pass or fail. They never hold a listing back.

InsightRuns onGrade A fromB fromC from
Test coverageeach codebase80%60%30%
History deptheach codebase24 months12 months6 months
Contributor spreadeach codebase10 people4 people2 people

Below the C mark the grade is D. With no figure to grade, the insight ends as Couldn't finish.

What a check can end as

ResultWhat it means
PassedIt found what it looks for.
FailedIt did not. The result says what to fix.
SkippedIt did not run: its price was not covered, a required check had already failed, the vendor chose Skip review, or the listing was submitted again, left review or had its text changed before it ran.
On our sideA service the check needs did not answer. Credits charged for it are given back; if the credits cannot be given back at once, the result says the refund is owed.
Couldn't finishThe check could not reach a verdict. A priced review keeps its charge, unless it ran out of attempts.
WaivedA required check that did not pass, waived by DataDrop with a note.

On the live site, a task's QA results, your credit balance and the AI reviews come from services DataDrop does not reach yet, so those checks end On our side there.

When the checks end

  • Every required check passed or was waived: the listing moves to Ready. Publish it from its page with Publish listing; see Manage a listing.
  • A required check failed: the listing moves to Needs attention, and your organization is told by a notification, A listing failed a check. Fix what the check names and submit the listing again: every check runs again.
  • A required check ended On our side or Couldn't finish: the listing stays in Review until DataDrop waives the check or rejects the listing.

DataDrop's operators see every listing the checks hold. An operator can waive a required check that did not pass, with a note, or reject the listing. A Secrets scan that found a credential is never waived: take the credential out of the text and submit again.

The listing's page does not show the check report in the web app yet. The vendor API's GET /v2/listings/{listing_id}/qc answers it, and so does the MCP tool get_listing_checks; see Using the API.

The verified badge

A listing holds the Verified badge when its latest review cleared without a waiver and every recommended check of it passed: none skipped, none on our side. Any change to what a buyer pays for (the listing's text, its assets, their prices or revisions) takes the badge away, and the next review decides it again.

The catalog's Verified only filter does not read the badge yet: it matches no listing.

On this page