Hani retweeted
Numbers talk. π
AgentOS in 84 days:
- 37.5K+ PyPI downloads
- 5,484 installs in 24h β our best day yet
- 2,592 commits, 66 releases
- 40 contributors, 996 merged PRs
More builders and users join every day. Thank you π§’
β Star us: github.com/use-agent-os/agenβ¦
The truth is, not many projects survive one growth wave after TGE. Most are one-wave memes, or a product with a pretty vision that slow rugs after that single pump.
At @useAgentOS, we build for the long term. We are open source. The team & community has impressive dev progress right now and keeps shipping. Token is still holding above $400k, and the holder base stays steady.
So if you want a project to hold long term, let me introduce you to robinhood:0x6eda83fc299c10d474068a7e69771c809bcbbba3.
Hani retweeted
ββ±ββββββββββββ²
β² Give us a star βοΈ β
β Every star helps. β
β Day by day better. β
β²ββββββββββββ±
Hani retweeted
π€ AgentOS v2026.9.10 Released
β Channel Reliability: correct Microsoft Teams routing, safer Telegram links, and reliable Discord commands
β Document Skills: preserve DOCX formatting, improve XLSX updates, and maintain PPTX structure
β MCP Reliability: prevent concurrent stdio read conflicts and bound email retries
β Runtime Operations: safer Windows upgrades, enforced browser session limits, and better Ollama diagnostics
β Performance & CLI: faster memory sync, optimized session checks, and `agentos --version`
---
Build AgentOS, earn @RobinhoodApp Tokenized Stocks.
β useagentos.dev/contributors
Hani retweeted
Within the last month of AgentOS build:
π£ 241 merged PRs
π’ 113 open PRs
β
242 closed issues
π 61 new issues
π¦ 24 authors, 377 commits to main
π 101,405 additions vs 6,855 deletions
Shoutout to every builder shipping on AgentOS. The agent runtime is getting sharper every single week, and it is all community-driven π«‘
18 contributors jumped in this week to make the AgentOS package stronger.
but i still see a few builders havenβt starred the repo yet. please drop a star β, itβs free π
Btw, reward by tokenized stocks really works.
Join us here: useagentos.dev/contributors
A friend once told me: people often think a chart needs a market maker, supply accumulation, or the right chain with enough liquidity to pump. But those are just secondary factors.
The real key is having a real product and a story that fits the current meta. That is the catalyst that brings in organic volume.
At the end of the day, organic volume is what actually drives the chart up.
Hani retweeted
10 code with $10 inference each airdrop to 10 lucky people who like, rt and tag 3 friends to below post π«‘
πͺ Big news: Inference Vouchers are live on OpenCAP.
π Giveaway: 10 code with $10 inference each will be airdrop to peoples who like, rt and tag 3 friends in this post.
Here is how it works. We issue the vouchers, you redeem them into your OpenCAP account, generate an API key, and boom, OpenCAP becomes your LLM gateway.
A quick note: each code has its own quota and can only be used once.
We are also open to collab with AI communities and agent builders to giveaway vouchers to your members. A simple way to onboard newcomers and let them experience AI hands-on.
AI is a must now, not optional. Use AI from today if you don't want to be left behind.
P.S. Don't forget that our inference can be sold on @AskSurplus if you don't use it.
#InferenceVoucher #OpenCAP
πͺ Big news: Inference Vouchers are live on OpenCAP.
π Giveaway: 10 code with $10 inference each will be airdrop to peoples who like, rt and tag 3 friends in this post.
Here is how it works. We issue the vouchers, you redeem them into your OpenCAP account, generate an API key, and boom, OpenCAP becomes your LLM gateway.
A quick note: each code has its own quota and can only be used once.
We are also open to collab with AI communities and agent builders to giveaway vouchers to your members. A simple way to onboard newcomers and let them experience AI hands-on.
AI is a must now, not optional. Use AI from today if you don't want to be left behind.
P.S. Don't forget that our inference can be sold on @AskSurplus if you don't use it.
#InferenceVoucher #OpenCAP
Hani retweeted
Robinhood CEO: @vladtenev on The Ice Coffee Hour: 2 hours on AI agents, agentic trading, and where finance is heading π§΅
The AI agent + agentic trading takeaways:
1/ Robinhood's MCP server lets AI agents (Claude, Codex) trade real money. Equities live, options added, crypto coming.
2/ Already 100k+ agentic accounts driving real volume. Not a demo. Actual trading.
3/ Funniest bit: agents sometimes refuse to trade. "Didn't feel like it." Agentic trading is so new it's not even in their training data.
4/ AI will disrupt active trading the most. Passive buy and hold? Barely touched. "I love my iPhone so I buy AAPL" behavior isn't going anywhere.
5/ Robinhood Chain was built AI-agent-first. Agents can hold, send, and swap tokenized stocks (Tesla, Nvidia) in 120+ countries.
Big picture: the interface to finance is shifting from apps to agents. The broker that ships the best agent rails wins.
P.S.
- This recap was transcribed and summarized by local agent running on AgentOS.
- People can install AgentOS then start trading on @RobinhoodApp via our built-in Robinhood MCP integration.
Check your usage for this week, and donβt miss our limited promo: get a 30% bonus when you deposit funds into OpenCAP, just one week.
Six contributors on the @useAgentOS board this week and it's just getting startedπͺ
To builders, remember to star and follow our repo, every star helpβοΈ
Hani retweeted
GM community,
Ignore the fud, we're happy to announce that now you can trade $MPMF on @BinanceWallet.
Check out here:
web3.binance.com/en/token/roβ¦
More pools more fun π₯³
Thrilled to announce that GLM-5.3 available on OpenCAP
We are now discounting 12% for this model.
GLM-5.3 by @Zai_org , released on August 14, 2026, keeps the same 743B-parameter base as GLM-5.2 but doubles down on scaled post-training RL, rocketing Terminal Bench 3.0 from 4.6 to 28.3 and DeepSWE v1.1 from 46.2 to 66.9.
In cybersecurity, it scores 84.5% on CyberGym, narrowly beating both Mythos 5 (~83.8%) and GPT-5.6 Sol (~83.6%) at vulnerability detection, while ExploitBench doubled from 24.4% to 54.4%.
The model also proved itself in the real world by finding 2,436 vulnerabilities across 269 open-source projects (1,097 rated critical/high), though it still trails top closed-source models on deep exploitation and some internal code benchmarks.
Overall, GLM-5.3 is now the strongest open-weight model for coding/agent work.
Give it a try now: capminal.ai/models
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense.
- Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model
- A major leap in cybersecurity, setting a new standard among open models
Tech Blog: z.ai/blog/glm-5.3
Hani retweeted
GM GM,
@pools_dot_fun ecosystem is on fire π₯
If you missed $sushicat, you still have a chance with $MPMF.
More pools more fun π«‘
Congrats to $102k volume today, here are something about $MPMF:
1/ I release a skill to launch token on @pools_dot_fun for AgentOS.
2/ I demo this skill to launch MPMF: nitter.cf/useAgentOS/status/2087β¦
3/ This token receive $50k vol just in 1 hour so I collected fee, buy back and burn all to support buyer: nitter.cf/useAgentOS/status/2087β¦
4/ This token continue to reach $90k volume so I continue to collect fee, build a website: morepoolsmore.fun , then paid DEX.
5/ Now $102k + volume π₯, if it's big enough, Iβm willing to cto it and keep supporting technical π«‘
More pools more fun.
Iβve collected the fees, added a bit of my own money, and paid the DEX, hoping it helps the trenches trading $MPMF.
The creator fee on @pools_dot_fun tokens is small, only 20% of 1% of trading volume. Most of the fees flow back into the ecosystem for daily buybacks of the top tokens. If the MPMF community grows and volume stays high, the protocol can buy it back.
More pools more fun π«‘
Volume now nearly $90k
I've just collect some more fee to build a community website for $MPMF, If community can grow large enough, Iβll look for someone to take cto.
Maybe I need @0xDeployer guide how to do cto @pools_dot_fun token π
More pools more fun.
morepoolsmore.fun
Hani retweeted
1/3 @pools_dot_fun Token Launcher skill is live on AgentOS latest version.
From preflight β deploy token β collect fee, run fully on your local.
Just need to set your private key to deploy, run fully on your local.
If you want to upload your token image, we have already integrated with @pinatacloud for this. Just need to import PINATA_JWT to env and run fully on your local.
C/c: @alexmccurryo @0xDeployer
Visual breakdown π
Thesis:
@useAgentOS is your personal AI agent that runs locally and helps you:
> Stop overpaying for top AI models with AgentOS Pilot Router, an on-device router that sends each request to the best-fit model
> Find any RWA contract address on @RobinhoodCrypto chain
> Agentic Trading out of the box via @RobinhoodApp MCP, @bankrbot skills, and @Capminal skills
> Auto-manage your @Uniswap V4 LP using the 'senior-unilp-manager' skill
And more....
How about your thesis ? Share it now on @fomo π
fomo.family/tokens/robinhoodβ¦
π€ senior-unilp-manager skill
Manage your @Uniswap V4 LPs directly on your local with AgentOS.
This skill has been bundled together with AgentOS v2026.8.2, no need to install.
---
Supported chains: Robinhood
Required env: RPC_ROBINHOOD_URL, UNIV4_LP_PRIVATE_KEY
Free RPC is still okay, but we recommend using paid RPC to avoid 429 errors because this skill uses many multicall queries.
UNIV4_LP_PRIVATE_KEY is required in case you want to manage your LP directly on AgentOS.
All env are secured on your local.
---
Check out the demo below:
> Query deeply LP positions of $AGENTOS
> Add LP one side specific range






































