Releases, at the top of the Know sub-nav, shows every publish by source id and how it landed with people and agents. It needs the Growth plan; below it the view still opens with its totals, and sample rows blurred beneath an upgrade prompt.
What a release is
A release is recorded whenever pages are published by source id: by the release:docs job that the continuous documentation setup adds to your CI, or by a direct call to schedule_pages_by_source_id. Each one keeps a run report of what the job found. A release takes the name its caller sends, such as a version tag; a call with no name gets a generated, date-based name. Until a release is recorded, the view says so and links to setting up docs CI with the /leed:docs-ci plugin skill.
The Releases item in the sub-nav carries the latest release’s name as its badge.
All releases
All releases (/know/releases/all) lists every release, newest first:
- Release to first read: how long each release took to get its first read, as a human median and an agent median. Pages published outside a release are left out.
- The release table: pages shipped, time to first read, and who was told (Email, Social, both, or Nobody told).
- Release process: Docs published by source id, Outside a release and Failed.
A Show switch narrows the list to All, Releases or Outside a release, and a Reads window switch picks 24 hours, 7 days or 30 days after each publish.
A release’s scorecard
Opening Releases shows the newest release; the row of recent releases across the top opens any other, and All releases at the end of it opens the full list. A scorecard has:
| Section | What it shows |
|---|---|
| Release run report | What the release job found when it ran, kept with the release: pages Matched, Scheduled, Already published and Failed, with the reason for each skip. Dry runs are kept too. |
| Headline tiles | Pages shipped, First read, reads in the chosen window after publish, and pages published Outside a release |
| What shipped, and who read it | Every page stamped with a commit in the release, with tabs All, Unread and Agents only |
| Reads by window | Reads in the 24 hours, 7 days or 30 days after publish |
| Adoption curve | Cumulative distinct readers of the release’s pages, by week since publish. It needs two weeks of data. |
| Who was told | Emails and shares that name a page of the release |
| Reach across customer accounts | Which customer accounts read the release. It needs account analytics and accounts marked as customers in Engage. |
Compare with sets the scorecard beside an earlier release at the same age; by default it is the previous release.
The scorecard takes the page type, readers, agents and account filters. It has no prior-period comparison of its own: the comparison is the other release.
Retention
Run reports older than your plan’s analytics history are left out. A release whose early days fall before that history has nothing to compare with.
Over MCP and the CLI
get_know_view reads one release with view: "release" and its release name, or every release with view: "releases". list_release_runs returns the run reports, and get_next_release_preview the latest dry run of the next release. From a terminal: leed know show releases, or leed know show release --filter release=<name>.
See Reading Know from MCP and the CLI and, for setting up the release job, continuous documentation. What the next release will publish is also on Knowledge health.