# About stalebot configuration In this last week I've seen a couple issues where stalebot was a problem. See https://github.com/decidim/decidim/issues/4667#issuecomment-1048161279 and https://github.com/decidim/decidim/issues/4667#issuecomment-1048161279 Sometime ago we thought that adding a bot that lowers the number of issues that we can focus was a good idea, but this could not be the brightest idea after seeing how this works in the real world. As @microstudi [summarized](https://github.com/decidim/decidim/issues/8909#issuecomment-1054113131) in one of those issues: > For respect to all the people that bother to create issues, I'd remove the stalebot completly and deal with old issues manually. There's lots of discussion about stalebot in the developers community, as an example there's [this HN submission](https://news.ycombinator.com/item?id=28998374). Our [configuration](https://github.com/decidim/decidim/blob/develop/.github/stale.yml) used to be more aggressive, now if it has a label/project/etc it won't be closed. But specially on issues created before these changes, we'd need to check all the closed issues by it, with labels "stale-issue" and "wontfix" to check: 1. Check that they can still be reproduced in the develop version (v0.27.0.dev at the moment of writing this). On those cases we should remove the [stale-issue](https://github.com/decidim/decidim/issues?page=1&q=is%3Aissue+sort%3Aupdated-desc+label%3Astale-issue+is%3Aclosed)/[wontfix](https://github.com/decidim/decidim/issues?q=is%3Aissue+sort%3Aupdated-desc+is%3Aclosed+label%3Awontfix) label and give them the 'bug' label. 2. If it's feature proposal (created previously to Metadecidim roadmap process dogfooding strategy), then we should invite the original creator to go to Metadecidim to create it. We can write a template for those cases. 3. We should also change the behaviour on closing issues. I think having a stale-issue label is good, and also that the bot leaves a message saying that this is indeed stale, so we can check the issue manually and see if it's still pertinent. What's a bad idea is to have an automatism for closing issues. For what I've read the solution would be to have "days-before-close: -1" so it doesn't close them. If anyone has any other idea/strategy/configuration, this would be the place to talk about it.