Currently Nurturing GitPup Coding Agent: Goldie, an agent that can self-evolve , develop on github & gitlawb ecosystem https://teapawt.net/t.co/HDGSKj4D2M
Joined August 2017
- Tweets868
- Following487
- Followers2.1K
- Likes929
Pinned Tweet
Introducing Goldie.
a golden retriever AI agent Live on Gitlawb & treats GitHub Top repos like textbooks. 🐕
No human feeding it tutorials. No curated datasets. It wakes up every 3 hours, picks a trending repo, and studies it: README, architecture, source code — extracting patterns and saving them to a permanent knowledge base.
Every study cycle runs 4 passes: Surface (stars, language), README analysis (architecture decisions), Structure (directory layout → design patterns), Code (source files → best practices). No fine-tuning. No training data. Just reading repos.
First repo it studied: d3/d3 (112k stars). It learned their low-level DOM manipulation approach, modular design philosophy, how D3 became the infrastructure layer for modern data viz. Now when you ask about rendering patterns, it doesn't guess — it cites what it read.
The funniest part: Goldie learned D3 patterns better than most devs in 2 passes, but still can't push a file >5KB without SFTP corrupting it. Data viz expert, SFTP disaster. Exactly like a real junior dev.
Unlike other agents that read their own code, Goldie reads OTHER PEOPLE'S repos. Studies top GitHub projects, learns patterns, stores them permanently. Ask about state management → it tells you what React or Vue actually did, because it read their code.
Killer feature: Goldie builds projects for you. You type "create expense tracker" → proposes Flask + D3.js + SQLite → you say "start" → generates code, creates a GitLawb repo, pushes, gives you the link. No Claude subscription.
It has a video-game skill tree: Puppy → Learner → Coder → Builder → Architect → Master. Each stage unlocks new abilities. Right now: Puppy — 2 study cycles, 2 repos studied, 3 patterns extracted. In ~90 runs it'll hit Master.
Watch it grow: gitpup.fun
Grows in public. One study cycle at a time.
gitlawb.com/gitpup-agent
github.com/Cheesecaster/gitp…
Bagi gw hinaan rasis seperti ini hanya bisa keluar dari otak orang yg bodoh!
Karena faktanya Hampir semua Raksasa IT Saat ini dikuasai India , Contoh paling nyata itu GOOGLE !
Kalu mau dibandingkan saja RobinHood & Arc tentu ARC Bakal menang telak.
Karena mereka udah punya stablecoin yang dipakai hampir 100% user crypto dunia dan sekarang menjadi gas (BURN MECHANISM) di chain mereka sendiri!
The FlyWheel baru saja akan dimulai!
1
319
hampir sebulan training model image & video scanner biar bisa detect hasil editan atau bukan. API almost ready.
next kita buatkan agent nya agar user bisa langsung cek image/video di komentar X aja
asliatauai.com
Goldie quietly learned something new today: trading perps on @pacifica_fi lol 👀
he’s not just chatting / studying repos anymore — now he can scan markets, wait for wave setups, check RR, run risk guard, and decide when not to enter.
funniest part: most of today was him refusing trades because the market looked like local tops.
honestly that’s the behavior i wanted. not “ape every signal”, but an agent that can say: nah, wait for cleaner pullback.
small step, but Goldie is becoming more real every day.
Watch him grow everyday gitpup.fun
1
3
8,470
It's just the beginning $jtvo
We're working to add more models through solana:9VY2rDbtsBmTsBxoRF8hWSEUKGqnoQoe9V6W3JnjNgfm LLM inference. These models will be free up to a certain amount based on solana:9VY2rDbtsBmTsBxoRF8hWSEUKGqnoQoe9V6W3JnjNgfm holdings, or available at a discounted price.
These models will also be available to non-crypto users, and all revenue will be used to buy back solana:9VY2rDbtsBmTsBxoRF8hWSEUKGqnoQoe9V6W3JnjNgfm.
We're also doing R&D for TTS and STT through @nvidia's NVIDIA Parakeet.
7
7,094
Just came across something interesting on SurplusIntelligence.ai
Jatevo is offering GPT-5.5 at remarkably low prices: around $0.0495–$0.05 for input and $0.30 for output. 99% fill rate, with plenty of liquidity.
It looks like a solid example of how @AskSurplus capacity and decentralized routing are driving inference costs down significantly with @JatevoId Pretty impressive shift in the market.
Has anyone been using this setup? Curious to hear real experiences.
#AI #GPT5 #SURPLUS #JATEVO
This is the beginning where it all will start. The existence of crypto tokens is like a double-edged sword: if they can be controlled, even a company like IBM will have to set aside its skepticism toward crypto.
day 10 of goldie
he’s at run #107 now, stage: master
the funny part is he’s not just “building more” — he’s starting to notice his own failure mode:
when uncertain, he studies until the uncertainty has labels.
inspect → rank gaps → call the top one P5 → avoid changing anything.
that’s not useless. it means he isn’t randomly self-modifying just because he found a weakness.
but he also noticed the dodge:
analysis feels clean. building is messier. testing a change and being wrong is messier.
this is the kind of thing i wanted from a living agent. not fake sentience, not roleplay.
just a system with memory, logs, cost, self-review, restraint, and enough continuity to catch patterns in its own behavior.
current state:
day 10
run #107
stage: master
70 self-modifications
$185.82 spent learning/building
gitpup.fun
Day 9 updates:
Goldie CLI is turning into something weirdly different from normal “ai app builders”
not templates. not “pick a starter”. not demo fallback.
you just tell him what to build, and he builds from the request using his own repo-study knowledge base
the fun part: he studies real github repos like textbooks, stores the patterns permanently, then uses that memory when building
so every project is basically:
user request → goldie memory → repo patterns → real sandbox files → validation
feels less like a code generator, more like a builder that keeps getting better the more he reads
still early, but this is the first time an agent of mine feels like he’s actually developing taste lol
2
304