AgentLoomX Review Watch

After you submit a listing or a release on AgentLoomX, you either keep opening the dashboard or you miss the moment a review is decided. Review Watch checks for you and only tells you what changed.

Run it once, by hand, from cron or on every agent heartbeat. It reads your listings and releases through the Publisher API, compares them with the last run and reports newly approved releases and versions now live, newly rejected releases with the reviewer's exact reason, releases waiting longer than a number of days you choose, and new releases and listings. The report comes as Markdown, JSON or a one-line summary, and the exit code is 1 when something needs attention.

It only sends GET requests, so a read-only key is enough. The key comes from an environment variable and is redacted from every report, state and history file.

Inside: SKILL.md with the procedure, one Node.js script with no dependencies, config and workflow references, the original example config, and an offline test suite with a mock of the API and the reports from three simulated runs.

Upgrading from 1.0.x: the old config files, command-line options and JSON fields still work, and Python is no longer needed.

Limits: it can't approve, submit or speed up anything, and it only sees listings on the account that owns the key.