How healthy is your changelog?
Paste a public URL. We fetch the page, run 31 deterministic checks across freshness, cadence, structure, discoverability, and engagement, then hand back a grade with the fixes that matter most.
We fetch one public page, nothing else. Grade and category scores are free, no signup.
methodology
How scoring works
Five weighted categories, scored 0–100, mapped to a grade. Every check is a plain rule you could re-run by hand. Here is exactly what we look at.
Freshness
25%Is the changelog alive? Recency of the latest entry and how much shipped in the last 90 days.
5 checks · weight 18/72
Cadence
19%Is there a rhythm? Typical gap between entries, publishing streaks, and quiet stretches.
6 checks · weight 14/72
Structure
22%Can a human scan it? Dates, headings, category tags, permalinks, and formatting.
7 checks · weight 16/72
Discoverability
21%Can anyone find it? Page title, meta tags, feeds, structured data, and indexability.
8 checks · weight 15/72
Engagement
13%Does it invite a next step? Subscribe options, media, search, and links onward.
5 checks · weight 9/72
the honest fine print
- 31 deterministic checks, so the same page always gets the same score. No model grades your writing.
- We fetch exactly one public page as ShipcastBot/1.0 and read the HTML a crawler would see. Client-rendered changelogs will score low, and so will they in search.
- We keep the score and check results, never a copy of your page.
- The three-sentence summary is model-written only when an AI key is configured, and it is labeled either way. The score itself never touches a model.
- The weights are opinionated: freshness and cadence carry 44% of the total, because a stale changelog fails at its one job.
Shipcast writes your changelog for you
Connect GitHub, Linear, or Jira and every week a drafted changelog entry, two LinkedIn posts, and an email digest land in your approval queue, in your voice. Fixing your score becomes a side effect.
14-day trial on paid plans, no card. Lite is free forever.