home-assistant/core: Worth your time?

Loading starter issues…

home-assistant/core: Worth your time? · Holt

how this was counted

what we read: 325 pull requests on GitHub, opened 14 Sep – 29 Sep 2026.
the team and outside contributors: 128 of them came from 24 people on the project's team and are left out. The team is everyone GitHub shows as an owner, member or collaborator, plus anyone the history shows merging, closing or formally reviewing other people's pull requests. Everyone else is an outside contributor, including people who come back.
bots: 11 pull requests opened by bots (dependency updates and the like) are left out. Only a maintainer's reply counts as a reply: not a bot, a CLA check, the author, or a passer-by.
drafts and spam: 46 pull requests from outside contributors were drafts or were marked as spam or invalid, so they aren't counted anywhere.
too recent to count: 73 pull requests from outside contributors were opened in the last 14 days, too recently to know how they will end, so they aren't counted yet.
closed by their authors: 2 pull requests from outside contributors were closed by the people who opened them.
what decided it: 49 pull requests from outside contributors were merged, by 31 different people, out of 67 attempts by 45 people. Of those who got a reply, half heard back within 15.7 hours. 8 attempts got no reply at all and weren't merged.
the rule: Only pull requests from outside contributors opened between 14 days and a year ago count, so each has had time for an answer and reflects how the project works now. “Worth your time” needs at least 2 of them merged, from at least 2 different people, with at least 3 in 10 people getting their first pull request merged (once 8 or more have tried), at least 1 in 10 merged, no more than half left without any reply, a typical first reply within 3 weeks, and people actually reviewing what gets merged. “Long shot” is a project that merges some outside work but misses one of those: fewer than 3 in 10 newcomers getting in, fewer than 1 in 10 merged, most left unanswered, very slow first replies, or only one person's work getting in. “Not worth your time” is an archived repository, a mirror, a project with nothing merged or pushed in 90 days, merges nobody reviews, fewer than 1 in 20 outside pull requests merged once at least 20 have been decided, or outside pull requests that are almost all ignored. “Personal project” is someone's own project or a small team's, like a hackathon entry, with nothing from outside ever merged. “A list, not code” is a list or a catalogue of entries, like links or package listings, where a merged entry isn't code work. Anything in between is “Not enough evidence”. These rules are fixed; no AI chooses the verdict.
what Holt couldn't check: No AI read the conversations for this report, so it doesn't say how maintainers talk to newcomers or what kind of project this is. The verdict and the numbers are counted straight from GitHub and don't need an AI.
where the advice comes from: Where the project asks you to say whether you used AI; Where a bot closes quiet pull requests

Worth your time.

Your first contribution

What happened to outsiders

Where newcomer work lands

Was this verdict right?

The evidence

What to do here to get merged