Moderate SAP Commerce reviews with four questions

01 · Role in the system

What Jev does here

For each review, one request carries four typed questions with probabilities: is it abusive, is it spam, does it contain personal data, is it about the product. The extension maps the answers to approve, reject, or leave pending - and anything Jev is not sure about stays pending for a person. Version 1 leads with measurement: dry-run mode judges reviews human moderators already approved or rejected and records Jev's decision beside theirs without changing anything, writing one summary line per language to the cronjob and server logs, so you measure on your own data before it changes anything. The integration is a plain Java extension - a minimal client posting to the System One endpoint over the JDK HttpClient, with integration tests covering the client and the review-moderation job.

02 · Control boundary

Where Jev sits

One four-question batch per review; deterministic code maps probabilities to moderation outcomes with a pending-by-default middle band; dry-run replays against human decisions before enforcement.

Code owns the loop, permissions, thresholds, validation, and side effects. Jev owns only the bounded judgments described above.

03 · Known limits

What this evidence does not prove

• An independent project, not made by or affiliated with TypeSafe or SAP, per its own README.
• Deploying means review text reaches the API; the dry-run mode exists precisely to measure that flow before enabling enforcement.
• Four narrow questions cover v1; other moderation needs are out of scope.

04 · Attribution

Public sources

This is a Community record: the project was published by a third-party community author.

• Emenowicz ↗Community · github · public · checked 2026-09-24