hailey
@hailey.at
engineering stuff, safety tools, and patterns
@labeler.hailey.at
@labeler.hailey.at
wonder how long itll take to backfill the follow graph
December 26, 2025 at 8:59 AM
wonder how long itll take to backfill the follow graph
my for you feed is fucking insane rn
December 26, 2025 at 7:50 AM
my for you feed is fucking insane rn
Reposted by hailey
kwargs is the nastiest programming language feature. it's just senselessly indecent
December 26, 2025 at 7:31 AM
kwargs is the nastiest programming language feature. it's just senselessly indecent
fatal error: robpike: reading email
goroutine 3182 [running]:
robpike/tasks.fatal({0x28952e0?, 0xc00691d9d8?})
runtime/panic.go:1056 +0x18
robpike/tasks.(*Tasks).readEmail(0xc006baf2a0, 0xffffffff)
robpike/email.go:204 +0x35
goroutine 3182 [running]:
robpike/tasks.fatal({0x28952e0?, 0xc00691d9d8?})
runtime/panic.go:1056 +0x18
robpike/tasks.(*Tasks).readEmail(0xc006baf2a0, 0xffffffff)
robpike/email.go:204 +0x35
December 26, 2025 at 5:16 AM
fatal error: robpike: reading email
goroutine 3182 [running]:
robpike/tasks.fatal({0x28952e0?, 0xc00691d9d8?})
runtime/panic.go:1056 +0x18
robpike/tasks.(*Tasks).readEmail(0xc006baf2a0, 0xffffffff)
robpike/email.go:204 +0x35
goroutine 3182 [running]:
robpike/tasks.fatal({0x28952e0?, 0xc00691d9d8?})
runtime/panic.go:1056 +0x18
robpike/tasks.(*Tasks).readEmail(0xc006baf2a0, 0xffffffff)
robpike/email.go:204 +0x35
Japanese cluster over the past six hours. The entirety of the lime green section at the bottom is a combination of "Japanese Artists" and "Japanese Bluesky Migration"
December 25, 2025 at 4:58 AM
Japanese cluster over the past six hours. The entirety of the lime green section at the bottom is a combination of "Japanese Artists" and "Japanese Bluesky Migration"
it’s so bad that one of the hai di lao waiters just knows us now.
doing hai di lao again twice in one week is so good
December 24, 2025 at 5:03 AM
it’s so bad that one of the hai di lao waiters just knows us now.
doing hai di lao again twice in one week is so good
December 24, 2025 at 3:28 AM
doing hai di lao again twice in one week is so good
added a "tap mode" to atkafka. i've found that it is much easier to work with kafka when working with services where you really do not want to lose events
Add Tap mode by haileyok · Pull Request #2 · haileyok/at-kafka
tap mode
add acks
update dockercompose with tap
update readme
github.com
December 23, 2025 at 10:31 PM
added a "tap mode" to atkafka. i've found that it is much easier to work with kafka when working with services where you really do not want to lose events
Reposted by hailey
is it time to re-up this
December 23, 2025 at 6:06 PM
is it time to re-up this
Reposted by hailey
login with wherever you get your brainworms
December 23, 2025 at 6:27 PM
login with wherever you get your brainworms
Reposted by hailey
gen z sysadmins be like rm -fr
December 23, 2025 at 3:17 PM
gen z sysadmins be like rm -fr
Because @labeler.hailey.at is fairly low stakes and is merely a community tool (not something we run in production), I've gone ahead and made the ruleset that drives it public. Hopefully serves as a good reference for writing Osprey rules.
GitHub - haileyok/atproto-ruleset: An Osprey ruleset made for moderating Bluesky and other ATProto applications
An Osprey ruleset made for moderating Bluesky and other ATProto applications - haileyok/atproto-ruleset
github.com
December 22, 2025 at 10:33 PM
Because @labeler.hailey.at is fairly low stakes and is merely a community tool (not something we run in production), I've gone ahead and made the ruleset that drives it public. Hopefully serves as a good reference for writing Osprey rules.
Reposted by hailey
I think this is far enough along to post about now
pegasus, a pds implementation in ocaml, ~feature parity with the bluesky pds (minus app passwords because those are gross)
repo includes various other atproto-relevant libraries in ocaml for interested sickos
tangled.org/futur.blue/p...
pegasus, a pds implementation in ocaml, ~feature parity with the bluesky pds (minus app passwords because those are gross)
repo includes various other atproto-relevant libraries in ocaml for interested sickos
tangled.org/futur.blue/p...
futur.blue/pegasus
objective categorical abstract machine language personal data server
tangled.org
December 22, 2025 at 7:12 PM
I think this is far enough along to post about now
pegasus, a pds implementation in ocaml, ~feature parity with the bluesky pds (minus app passwords because those are gross)
repo includes various other atproto-relevant libraries in ocaml for interested sickos
tangled.org/futur.blue/p...
pegasus, a pds implementation in ocaml, ~feature parity with the bluesky pds (minus app passwords because those are gross)
repo includes various other atproto-relevant libraries in ocaml for interested sickos
tangled.org/futur.blue/p...
Reposted by hailey
“why are all your posts on here about this”
idk maybe it’s because i got dogpiled with hundreds of people informing me that LLMs are markov chains and tons of crude insults?
the discourse on here for AI is fucking bleak
idk maybe it’s because i got dogpiled with hundreds of people informing me that LLMs are markov chains and tons of crude insults?
the discourse on here for AI is fucking bleak
yeah basically you can't really use the site for anything remotely to do with ai unless you're an extremophile. there's a reason why most of the people who talk about ai the most are Combat Posters, anyone who does not thrive on conflict simply stops bothering to post at some point
Deleted a post that was going viral and bringing an undesirable amount of attention from anonymous accounts. But, I'm going to be honest, the toxic folks here have won. So much of the usage of the site is about avoiding their attention. I'll just use this site to get book recs going forward
December 22, 2025 at 2:10 PM
“why are all your posts on here about this”
idk maybe it’s because i got dogpiled with hundreds of people informing me that LLMs are markov chains and tons of crude insults?
the discourse on here for AI is fucking bleak
idk maybe it’s because i got dogpiled with hundreds of people informing me that LLMs are markov chains and tons of crude insults?
the discourse on here for AI is fucking bleak
Reposted by hailey
Tis the season for open source governance :)
We're honored and excited to have some incredible people on the board for @roost.tools I look forward to working with these fantastic powerhouses to change the landscape of safety technology through open infrastructure and technology.
We're honored and excited to have some incredible people on the board for @roost.tools I look forward to working with these fantastic powerhouses to change the landscape of safety technology through open infrastructure and technology.
We’re proud to share an important milestone for ROOST: the formation of our inaugural Board of Directors! This Board brings together leaders across open source, technology, public policy, civil society, and national security.
roost.tools/blog/roost-w...
roost.tools/blog/roost-w...
ROOST Welcomes Inaugural Board of Directors to Guide Mission and Long-Term Governance
Robust Open Online Safety Tools or ROOST is a new non-profit entity designed to address the urgent need for accessible, high-quality safety tools in the rapidly evolving digital landscape.
roost.tools
December 22, 2025 at 3:56 PM
Tis the season for open source governance :)
We're honored and excited to have some incredible people on the board for @roost.tools I look forward to working with these fantastic powerhouses to change the landscape of safety technology through open infrastructure and technology.
We're honored and excited to have some incredible people on the board for @roost.tools I look forward to working with these fantastic powerhouses to change the landscape of safety technology through open infrastructure and technology.
Reposted by hailey
took a peek at the groupings of posts on bluesky over the past 48 hours
December 22, 2025 at 6:02 AM
took a peek at the groupings of posts on bluesky over the past 48 hours
istg my account is perpetually stuck in the "AI Ethics" cluster on here :/
December 22, 2025 at 6:13 AM
istg my account is perpetually stuck in the "AI Ethics" cluster on here :/
took a peek at the groupings of posts on bluesky over the past 48 hours
December 22, 2025 at 6:02 AM
took a peek at the groupings of posts on bluesky over the past 48 hours
who is ready for #GamerGate2, this time doing investigations and writeups about suspected use of AI...
yeah this sucks
That pretty much guarantees no one will admit to using any of these tools or practices again while disgruntled ex employees make wild accusations. Gamers are going to turn to AI investigating everything
Embark taking strays, Larian, too
Crappy postal game but who cares about postal
That pretty much guarantees no one will admit to using any of these tools or practices again while disgruntled ex employees make wild accusations. Gamers are going to turn to AI investigating everything
Embark taking strays, Larian, too
Crappy postal game but who cares about postal
December 22, 2025 at 5:18 AM
who is ready for #GamerGate2, this time doing investigations and writeups about suspected use of AI...
this is beyond silly lol
December 22, 2025 at 5:07 AM
this is beyond silly lol
a sample of what commonly happens with coordinated spam networks. we see that these accounts are all starting up with similar bios and all posting some sort of political/news content
December 21, 2025 at 10:42 PM
a sample of what commonly happens with coordinated spam networks. we see that these accounts are all starting up with similar bios and all posting some sort of political/news content
Reposted by hailey
this site never fails to find the coolest people who talk about the coolest things and yell at them for "believing" things that they never actually stated they believe. from sports folks to AI researchers to shit posters, it never fails.
December 21, 2025 at 9:28 PM
this site never fails to find the coolest people who talk about the coolest things and yell at them for "believing" things that they never actually stated they believe. from sports folks to AI researchers to shit posters, it never fails.
time to add some labels to @labeler.hailey.at for this
December 21, 2025 at 8:43 PM
time to add some labels to @labeler.hailey.at for this
this btw if why i am bullish on stuff like @letta.com. you can't wait years for stuff to show up in training data, nor are you going to do the training yourself. i should probably give letta code a shot.
not enough uv in the training corpus!!!
December 21, 2025 at 8:34 PM
this btw if why i am bullish on stuff like @letta.com. you can't wait years for stuff to show up in training data, nor are you going to do the training yourself. i should probably give letta code a shot.