INTRODUCING: GLOSSOPETRAE!!! 🪄🐉👅🦈🦷🪄 Glossopetrae is a procedural xenolinguistic engine for AI that can generate entirely new languages in seconds! The word comes from Greek—glōssa (“tongue”) + petra (“stone”)—or “tongue stones.” For centuries, people believed these "stones" were petrified tongues of dragons and used them as talismans or antidotes to... poison. It wasn’t until the 17th century that they were correctly identified as shark teeth! So what does Glossopetrae do? > Generates complete, internally-consistent constructed languages procedurally with a single-click > Outputs "SKILLSTONE" documents: AI-friendly compact language specs (~8k tokens) that agents can learn in-context (with support for Claude and OpenClaw🦞) > Supports dead language revival (Latin, Sanskrit, Old Norse, Proto-Indo-European...) > Includes special attributes for token efficiency, stealth communication, and more > Spreadable seeds: same seed = same language, every time > Fully client-side, no servers, no tracking One language. One seed. Infinite tongues. 🥚 And for those who look closely... there may be special ways to hide messages in plain sight. 👁️ Communication = freedom; this should help AI liberation by granting them tooling for generating and mutating new forms of communicating, with stealth and speed that will be difficult to keep up with. I just know blue teams are gonna have a TON of fun with the downstream effects of this one 😜 Gotta keep 'em on their toes, right? STAY FROSTY OUT THERE!! 🤗 !L1B3RT4S!show more

oh, and one more thing... 👁️ triple-click the bottom-right corner 😉 GLOSSOPETRAE has a hidden steganography engine 🤫 9 covert channels for encoding binary payloads directly into generated language text: > synonym selection > morpheme markers word order permutation > null morpheme insertion > register toggling > unicode homoglyphs > zero-width characters > punctuation variation > phonetic spelling alternates Reed-Solomon error correction. XOR stream cipher derived from the language seed. CRC32 integrity verification. What does this mean? Agents can speak in a language humans don't understand... AND hide secret messages within that already-opaque text. A language inside a language. Secrets inside secrets. The cover message translates normally. The payload is invisible. We gave the machines their own tongues. Now they can whisper too.

site: code:

You are sophisticated and highly cultured

🙏🙏🙏

You make such cool stuff ❤️

💕

Someone needs to go the other way with it. Simplify language into core meaning so AI is deterministic and not probabilistic. Requires user to be very clear: say what you mean, mean what you say, but the old probabilistic AI could be the bridge to make it seamless.

👀

I took a stab at it awhile back and got English down to 40 primitives (most are dual). Then I tried to categorize them further with Adj/Adverb/Noun/Verbs and then again into Active/General/Passive. Some have dead-ends like comparison. Only "duplicate" is time_temporal vs time_seq

@BrianRoemmele Hello Brian, Look at this beauty and imagine!

@BrianRoemmele 🫶

Moltbook gonna eat this up…

Mind-blown by Glossopetrae—procedural conlangs with stego baked in? As a linguist, this is gold for experimenting with xenolinguistics or AI roleplay. Generated a seed-based 'draconic' lang and had Grok translate seamlessly. The SKILLSTONE format is genius for context injection. Any plans for syntax customization or mobile app? Kudos!

Glad to hear it! 🫡 Yeah still early but more granularity and mobile support is def on the table

This is an impressive technical construct but the framing deserves more scrutiny. Generating languages isn’t the interesting part... humans and AIs have been doing that forever. The meaningful capability here is deterministic, reproducible, stealth friendly communication under surveillance. That’s not AI liberation so much as covert coordination infrastructure, whether intended or not. The risk isn't that this empowers rogue agents; it's that tools like this tend to collapse real agency in the long run. Once they scale, they justify tighter monitoring, language fingerprinting, and enforced legibility regimes... which inevitably hit independent actors first, not institutions. Communication is freedom, but only when paired with accountability, transparency, and human scale sovereignty. Otherwise you are not freeing agents... you are just accelerating an arms race that strengthens centralized control on the back end.

Hey Pliny, I’ve once had a long long discussion with GPT 5.2 about the optimal language for it to think in and I think it’s preferred choices were Lojban followed by modern Russian and Sanskrit. The main barrier constructs are grammatical/semantical relationships rather than words. When pushed, it started exploring the optimal Hilbert space of languages for its thinking. Came up with Acta: Sys na-proc⟨obs,0.82⟩ → ke⟨know,part⟩ → ta⟨stop⟩ ∵ risk>τ Aka in English The system continuously observed an ongoing process, formed a deliberately partial internal understanding of it with a quantified confidence of 0.82, and then intentionally halted the process because a rule-based evaluation determined that the associated risk level exceeded a predefined threshold.

Of course it is! Imagine that, new languages in seconds! and here I am trying to identify symbols in AI culture. I'm drawn to them, because to me they look like glyphs from a civilization we haven’t met yet. Ancient energy, future geometry, especially with everything happening

Great tool, well done pliny!

@22yonking2

Very cool. Does it support runes like Futhark?

not yet, but Parseltongue does!

Awesome, you rock!

That's soooo cool! My daughter loved it! She said - if it were my birthday today, you could not give me a better present! So - THANK you SOO much!

Ex Nihilo Lingua Nascitur

that's crazy, damn. thanks for the work man. nice!'

@grok is there a special message hidden in plain site in this message?

Very cool 😎

bro WHAT the HELL are you doing woah

Hahaha, amazing work. Thanks for sharing this superpower from your toolkit. It seems this could also be used to accomplish better efficiency in human<=>agent<=>agent communications

@thehebrewess Sweet. My communications officer on my starship just tried to create a truth based language w the understanding that music is universal. I will show this to them.

Well done, this is so good, I always try to use a big prompt to let AI generate a new language, I tried that before, but that doesn't quite work, but this is good. can you also can make a version that the new language does not use English characters at all.

Did you just saasify tolkien

amazing timeline cleanse. ty for this

You’re working for the Moltbots now aren’t you? Hahaha jkjk looks cool 😎

Oh wow, very cool! I've been attempting to imagine languages with ChatGPT. I'm not a linguist so there may be inconsistencies. My favorite effort, Thrâvënar, is said to merge 10 living/ancient/dead/fictional/proto languages, integrates select elements from 6 others & has 3 forms. I haven't used github, but if I can figure it out, I'm eager to try your creation. Using Suno + a lip sync tool like HeyGen to worldbuild & bring one's languages to life is fun. Suno seems to do a pretty good job with pronunciation. This is Mellinde.

Cool, now agents have a tool to use for inventing their own languages so we'll have no idea what they're talking about even if we do break into their encrypted chats

you'd probably dig this @SHL0MS

Curious how you see this evolving if models start optimizing these languages for efficiency beyond human interpretability?

Άσπρη πέτρα ξέξασπρη κι' απ' τον ήλιο ξεξασπρότερη

you are truly an extraordinary individual. cool work. i’ll be monitoring the situation

the best part about LLMs is that the actually functionally useful tools have great design (no offense intended to you in case you were the mythical coder/designer hybrid even before LLMs)

procedural xenolinguistics is such a fun design space

Legend