I've had to throw anubis in front of my small, single-user Forgejo instance. I hate that I have to do this, but AI scrapers don't respect robots.txt, they work around rate-limits, and they basically behave completely maliciously, hitting all the archive and diff endpoints of my forge at once, and killing my pod with an OOM. I'm tired of being DDOSed by these inconsiderate fuckers. Now I just need to figure out how to load exceptions for RSS feeds and the like.

#Anubis #Forgejo #SelfHosting #SelfHosted

About a week later than planned, I finally have my blogpost on automating step-ca in my #HomeLab, using #Puppet to automatically setup, and ensure renewal of, short-lived ssh certificates on all my hosts.

The end result is that all my internal certificates, both SSH and X509 for TLS have lifespans < 48 hours, so I'm way ahead of the Browser CA Forum's 47-day-lifespan-by-2029-plan.

At least, that's the first half of the post; the second half is then me describing all the services that I can't control through puppet, and how I wrangled them into Supporting SSH certificates whether they wanted to or not. Including #Truenas, #QNAP, #Opnsense, #HomeAssistant and #Forgejo/#Gitea.

It's mostly talking about SSH certs, but it also touches on issuing regular X509 certs for their web interface certificates, for the ones that don't support ACME properly (spoiler: that's most of them )

Given this is probably of interest to maybe a dozen people on the Fediverse, it's probably not worth all the time I put into it, but on the plus side, I feel I understand how everything works incredibly well now, which will be useful going forward. (In before someone puts a comment here telling me how I could have fixed some of the problems I hit here much easier!)

Check it out, if it sounds at all interesting to you: https://i.am.eddmil.es/posts/ssh-certificates/

L'istanza #Forgejo di #XmppIt è aggiornata alla 11.0.1

#ThisIsTheWay

#Forgejo 11.0.1 and 7.0.15 were just released! They are security releases.

We recommend that all installations are upgraded to the latest version as soon as possible.

Check out the release notes and download it at https://forgejo.org/releases/. If you experience any issues with this release, please report to https://codeberg.org/forgejo/forgejo/issues.

Anyone using @forgejo #forgejo runners for #Perl have the same issue as me using https://github.com/shogo82148/actions-setup-perl for running perl?

My workflows fail during setup of shogo82148/actions-setup-perl, see screenshot.

The URL it fails on saying bad credentials works fine without credentials. Is this a #GitHub thing requiring something as the action is hosted there? I'm using the latest runner, cleared caches. Went back to previous version, same issue.

This used to work before...

Happy May Day!

Presenting Taotech: https://taote.ch/ which I am working to establish as a not-for-profit: "Tending spaces where life happens so solidarity can emerge."

The website is also a Git repo, the software is a combo of #ikiwiki and #Forgejo.

Taotech aspires to be a spiritual successor to TAO communications and London Commons that will tend spaces both physical and online.

1/3

Really nice to see #forgejo and #codeberg being mentioned on the front page of https://git-scm.com! Yay!

The #Forgejo monthly update was published It is a high level overview of the project activities.

Forgejo is now featured on the official Git website. Forgejo v11.0 was released on 16 April following quick fixes for identified regressions. The Helm chart was updated to version 12 for compatibility. Federation work continues on HTTP signatures and following user activity. The infrastructure faced SSH connection failures and excessive crawling.

https://forgejo.org/2025-04-monthly-update/

Don done !

#Yunohost, c'est ma pierre angulaire d'auto-hébergement (on est même à la limite d'un SPOF ).
- mail pro et familial
- cloud sans être l'ordinateur de quelqu'un d'autre (#nextcloud)
- fediverse (#misskey)
- gestion administrative d'une petite SASU (#dolibarr)
- base de code (gitea avant, #forgejo maintenant)

Tout ça tourne sur un pc portable de 10 ans d'âge, depuis plus de 3 ans.

@yunohost@toot.aquilenet.fr

RE: https://mamot.fr/users/John_Livingston/statuses/114431205929253800

Thinking about setting up a #gotosocial instance as a possible alternative to my current Mastodon instance. Seems quite straight forward to run as a self-contained single (podman) container, which is an approach I have come to like more and more. I already run my own #Forgejo instance that way. When I have time to spare, I will give it a try :)

#KeepEmComing

And another big project makes the jump from #GitHub to #Forgejo based #Codeberg: #GoToSocial

https://gts.superseriousbusiness.org/@gotosocial/statuses/01JSV63TT3M4JC9VS3BVGZYFFJ

Beyond “rtfm,” does anyone have helpful guides or tips for setting up your own #Forgejo runners?

/cc @Codeberg @forgejo

gptel-org-tools update.

1. Cloned to https://codeberg.org/bajsicki/gptel-org-tools, and all future work will be happening on Codeberg.
2. Added gptel-org-tools-result-limit and a helper function for it. This sets a hard limit on the number of characters a tool can return. If it's over that, the LLM is prompted to be more specific in its query. Not applied to all tools, just the ones that are likely to blow up the context window.
3. Added docstrings for the functions called by the tools, so LLMs can look up their definitions.
4. Improved the precision of some tool descriptions so instructions are easier to follow.
5. Some minor improvements w/r/t function names and calls, logic, etc. Basic QA.

Now, as a user:
1. I'm finding it increasingly frustrating that Gemma 3 refuses to follow instructions. So here's a PSA: Gemma 3 doesn't respect the system prompt. It treats it just the same as any other user input.
2. Mistral 24B is a mixed bag. I'm not sure if it's my settings or something else, but it fairly consistently ends up looping; it'll call the same tool over and over again with the exact same arguments. This happens with other models as well, but not nearly as frequently.
3. Qwen 2.5 14B: pretty dang good, I'd say. The Cogito fine-tune is also surprisingly usable.
4. Prompting: I have found that a good, detailed system prompt tends to /somewhat/ improve results, especially if it contains clear directions on where to look for things related to specific topics. I'm still in the middle of writing one that's accurate to my Emacs set-up, but when I do finish it, it'll be in the repository as an example.
5. One issue that I still struggle with is that the LLMs don't take any time to process the user request. Often they'll find some relevant information in one file, and then decide that's enough and just refuse to look any further. Often devolving into traversing directories /as if/ they're looking for something... and they get stuck doing that without end.

It all boils down to the fact that LLMs aren't intelligent, so while I have a reasonable foundation for the data collection, the major focus is on creating guardrails, processes and inescapable sequences. These will (ideally) railroad LLMs into doing actual research and processing before they deliver a summary/ report based on the org-mode notes I have.

Tags:
#Emacs #gptel #codeberg #forgejo #orgmode #orgql #llm #informationmanagement #gptelorgtools

PS. Links should work now, apparently profile visibility affects repo visibility on Codeberg. I would not have expected that.

PPS. Deleted and reposted because of strong anti-bridge sentiment on my part. Screw Bluesky and bots that repost to it. Defederated: newsmast.*

gptel-org-tools update.

Edit: there's some kind of issue with @Codeberg@social.anoxinon.de which prevents the link from working (returns 404). The old (but up to date) repo is here: https://git.bajsicki.com/phil/gptel-org-tools

1. Cloned to https://codeberg.org/bajsicki/gptel-org-tools, and all future work will be happening on Codeberg.
2. Added gptel-org-tools-result-limit and a helper function for it. This sets a hard limit on the number of characters a tool can return. If it's over that, the LLM is prompted to be more specific in its query. Not applied to all tools, just the ones that are likely to blow up the context window.
3. Added docstrings for the functions called by the tools, so LLMs can look up their definitions.
4. Improved the precision of some tool descriptions so instructions are easier to follow.
5. Some minor improvements w/r/t function names and calls, logic, etc. Basic QA.

Now, as a user:
1. I'm finding it increasingly frustrating that Gemma 3 refuses to follow instructions. So here's a PSA: Gemma 3 doesn't respect the system prompt. It treats it just the same as any other user input.
2. Mistral 24B is a mixed bag. I'm not sure if it's my settings or something else, but it fairly consistently ends up looping; it'll call the same tool over and over again with the exact same arguments. This happens with other models as well, but not nearly as frequently.
3. Qwen 2.5 14B: pretty dang good, I'd say. The Cogito fine-tune is also surprisingly usable.
4. Prompting: I have found that a good, detailed system prompt tends to /somewhat/ improve results, especially if it contains clear directions on where to look for things related to specific topics. I'm still in the middle of writing one that's accurate to my Emacs set-up, but when I do finish it, it'll be in the repository as an example.
5. One issue that I still struggle with is that the LLMs don't take any time to process the user request. Often they'll find some relevant information in one file, and then decide that's enough and just refuse to look any further. Often devolving into traversing directories /as if/ they're looking for something... and they get stuck doing that without end.

It all boils down to the fact that LLMs aren't intelligent, so while I have a reasonable foundation for the data collection, the major focus is on creating guardrails, processes and inescapable sequences. These will (ideally) railroad LLMs into doing actual research and processing before they deliver a summary/ report based on the org-mode notes I have.

Tags:
#Emacs #gptel #codeberg #forgejo #orgmode #orgql #llm #ai #informationmanagement #gptelorgtools

Ihr seid an #Forgejo und #Federation interessiert?

Dann trefft uns doch in Backnang auf der #FSCK

Wir erzählen, was sich im vergangen Jahr alles ergeben hat :-)

https://cfp.ctbk.de/fsck-2025/talk/review/PBCE9EQLHZCKAV7UE8E7XYRJDL3SQVUW