Nerd sniped me into researching this a bit
But Iβm the Kubernetes container guy. Nix is out of my expertise lol.
@evan.jarrett.net
DevOps Engineer. Working on @atcr.io #kubernetes #homelab enthusiast Husband of @jarrett.mom North TX π
Nerd sniped me into researching this a bit
But Iβm the Kubernetes container guy. Nix is out of my expertise lol.
This was my last year, Iβm down to spam, and a few accounts that wonβt let me change emails without contacting support.
Still obviously use it for YouTube. I pretty much avoid places that do not provide email based sign in now. Set up wildcard aliases so you can marcospizza@mycoolemail.com
screenshot of workflow logs from tangled.org
β οΈ did you just echo your API_KEY in your workflow?
spindles will now automatically erase secrets from logs! rest assured, your API key will not be beamed to the world.
big thanks again, to @evan.jarrett.net for the contribution!
4/7
screenshot of a tangled workflow file
β³οΈ we now inject common env vars into workflows β³οΈ
no more munging git output, just use TANGLED_COMMIT_SHA or TANGLED_REF_NAME instead!
see the full list of variables on the docs: docs.tangled.org/spindles.htm...
big thanks to @evan.jarrett.net for the proposal & the implementation!
3/7
Dpop demon hunters
18.01.2026 03:35 β π 20 π 5 π¬ 0 π 0Waiting for the leaflet.
17.01.2026 01:02 β π 3 π 0 π¬ 1 π 0I have a similar problem with building a docker credential helper. I currently do confidential client flow on the appview, and create jwts for docker. I think ideally id want to have the helper be a conf client. That way keys for it donβt leave your machine.
16.01.2026 15:25 β π 1 π 0 π¬ 0 π 0Yeah so this made me look deeper into how I (and claude) am using indigo... turns out half of it is just raw http requests.
15.01.2026 04:01 β π 1 π 0 π¬ 0 π 0I was about to say I hadn't ran into this, but I looked and I guess I gave up and just did a GET on the xrpc directly instead of using indigo lol.
15.01.2026 03:41 β π 2 π 0 π¬ 1 π 0I would be totally down with those banners as long as donating actually removed them.
dev.pdsls.donated:true
So after a quick look, basically in one spot I was ingesting records where I shouldnβt have which partially created some appview DB entries.
And indigo has a way to verify the records, but I wasnβt using it because didnβt think I needed it until nowβ¦
I'm clearly seeing this DID in the logs, and its acting like a profile record partially exists for it. could be from before they suffixed the records.
Its not breaking anything, but something is still getting ingested and so its just now something I have to review.
I also can't obviously prevent you, or others from making broken records. I'm clearly going to have to be better about validating what I'm ingesting now.
12.01.2026 18:55 β π 2 π 0 π¬ 1 π 0i don't really care if you re-publish in lexicon.schema although it seems redundant.
But creating metadata records on our namespaces can cause problems when we try and ingest our records.
you could have created store.lexicon.status.io.atcr.* if you wanted individual records for each.
I wish you wouldn't have created random records on our NSIDs for this and instead prefixed them with store.lexicon.*
For example, what happens if I now create an io.atcr.log schema? does it collide with the record you created in the PDS already?
On a related note: pdsls.dev/at://did:web...
Wait this is silly the whole point of publishing them is for them to be discovered lol.
Noticed what seem to be xrpc endpoints in this list too? I didnβt realize it was standard to publish those
bsky.app/profile/evan...
07.01.2026 03:22 β π 1 π 0 π¬ 0 π 0Pagination bug :/ apparently when you have over 50 users you need to loop more, who would have thought.
06.01.2026 15:57 β π 1 π 0 π¬ 1 π 0Have you tried just hitting the breakers and seeing which circuit itβs on? Might help you figure out where itβs powered from.
If itβs a timer, it should be accessible somewhere.
I had thought of this too. but was thinking they could have turned into tennis balls
05.01.2026 19:03 β π 2 π 0 π¬ 0 π 0Should be able to:
git config βlocal user.email
I now deployed what should hopefully be the fix for this.
03.01.2026 23:41 β π 1 π 0 π¬ 1 π 0Merged.
Reminder to set your email address used for git in tangled settings so that commits are linked with your account.
I didnt deploy, i just tried clearing the cache.
The tldr is a race condition in the permission check causes you to get backed-off.
But you hit the race condition again.
Iβm surprised i havenβt hit this issue in testing. but @besaid.zone hit this yesterday too.
Should be able to try again now.
03.01.2026 21:44 β π 0 π 0 π¬ 1 π 0My initial thought is they all probably use to be brighter, and they are all dim now with age and weatherβ¦
Yours was just the first to go out.
Whats the point of a light that is super dim?
Ok I know they gave us like a week of bonus time, but i swear Iβm hitting the limit more the last 2 days than i ever had
02.01.2026 23:19 β π 3 π 0 π¬ 0 π 1Under βGot any good client metadata examplesβ realized i wasnβt specific until as i pressed reply
02.01.2026 22:27 β π 2 π 0 π¬ 1 π 0The link to leaflet seems to be missing the h in https.
Also I have an example that uses permission sets atcr.io/oauth-client...
Classic Claude still wanting to use βFederated with the Bluesky networkβ
But this an awesome idea. Iβll probably take it for a spin this week