Pinned Tweet
🔥 More Volatility = More Fees = More LP Rewards!
The chAGENTOS-WETH LP turns price swings into potential rewards.
⚡ Wrap $AGENTOS into chAGENTOS
⚡ Price gaps create arb opportunities
⚡ Arbs rebalance pools and generate trading fees
⚡ LPs earn fees + $CHOP rewards, boosting APR
🔁 More volatility → More arbs → More fees → Deeper liquidity → More volume
Why fear volatility when your LP can farm it ? 👀
AgentOS is now partnered with @ChopLogsApp to set up the chAGENTOS log on Chop 🔥
More liquidity and a fresh use case for $AGENTOS.
Takes about 2 minutes:
> Wrap your $AGENTOS into $chAGENTOS
> Add LP chAGENTOS-WETH
> Start farming, reward paid by $CHOP token
Let's go: choplogs.app/logs/0xa6308e0b…
P.S: we are building Chop skill for AgentOS 💪
An update on CAPU and daily Inference Credit
Effective 2026/10/27, 00:00 UTC, we're pausing $CAPU minting and CAPU staking for daily Inference Credit on Capminal.
We want to be direct about why.
The $1/day of inference per staked CAPU runs on a discounted inference supply. That supply has run into problems on the provider side. Right now we don't have a replacement that keeps the program's cost sustainable. We'd rather pause now than run something that breaks later.
Here's exactly what changes and what doesn't.
What's paused:
• Minting new CAPU
• Staking CAPU
• The $1/day Inference Credit per staked CAPU
That last one is the only thing on OpenCAP that stops. Everything else keeps running.
What still works:
1. You can fully exit at your own rate.
> Unstake CAPU (1-day cooldown) → burn CAPU to unlock your sCAP → unstake sCAP (7-day unbonding) → receive your CAP.
> Burning returns sCAP at your original lock ratio, not today's mint price, and the pause doesn't change that. Your principal is right where you left it.
2. CAP staking and $CAP rewards continue as normal.
Stake CAP, hold sCAP, earn CAP rewards and claim them anytime. Stakers still get monthly reward by CAP token.
3. The OpenCAP LLM Gateway stays fully live.
If you top up USDC on OpenCAP, nothing changes for you. It's the same gateway and the same API. Pay-as-you-go inference isn't affected.
In one line: the only thing we're switching off is the daily $1 inference allowance per staked CAPU. The gateway, CAP staking, CAP rewards, and your ability to get your CAP back all stay on.
If you hold staked CAPU today and want your CAP back, the path is:
> Unstake CAPU → wait 1 day → Burn CAPU → Unstake sCAP → wait 7 days → Receive CAP
> Or just sell CAPU on open market (take care of slippage)
Looking back: CAPU staking has been live for 6 months. In that time it delivered $46.8k of inference to our users, paid for by the protocol, with no one selling a single CAP to get it. We're proud of that number. It showed that "stake capital, mint compute" works in practice, not just in a whitepaper.
We know many of you minted CAPU specifically for the daily credit, and this isn't the update anyone wanted. We still believe compute as an asset is the right direction. But the economics have to work on the supply side too. We'll keep looking for an inference source and a cost structure that make the model sustainable. When we have one we can stand behind, we'll share an update. Until then, we won't put a date on it.
Thanks for building with us. 🫡
👉 An Update on Capminal’s Agentic Wallet:
Today we’re announcing a difficult but necessary decision: Capminal is winding down its agentic wallet product.
Over the past year we set out to build an AI agent that could hold funds, execute swaps, run DCA and TWAP strategies, and act autonomously on-chain on behalf of its user. We shipped a lot of it.
What we did not find was product-market fit, the usage never reached the depth or retention that would justify continuing to carry a product of this complexity and risk surface.
👉 What Capminal continues to operate:
Capminal remains focused on two lines of business going forward:
■ OpenCAP LLM Gateway — our inference gateway and developer-facing API.
■ Virtuals ACP services — our agent-to-agent commerce offerings on the Agent Commerce Protocol.
Both are actively maintained and continue to receive engineering investment.
👉 Timeline and what you need to do:
We will keep the agentic wallet infrastructure running for one month from this announcement. During that window:
■ Move your funds out of your Capminal agentic wallet, starting today. Do not wait until the final days.
■ If your Capminal wallet is the owner of any token you deployed through us, you can transfer owner to external wallet you control.
The hard deadline for both actions is 25 November 2026.
After that date, the agentic wallet, the Captain Dackie account on X, and the Telegram client will all be shut down. Any assets or ownership rights still held by a Capminal agentic wallet after the deadline may become permanently inaccessible.
Thank you to everyone who tested, used, and gave us feedback on this product. We’re carrying those lessons into what we build next.
Dackie retweeted
Every day, AgentOS grows a little more. Behind the numbers is a community of builders who keep showing up, committing code, and shipping packages that get better over time.
Join us and start building in public now:
github.com/use-agent-os/agen…
The ticker is robinhood:0x6eda83fc299c10d474068a7e69771c809bcbbba3
Dackie retweeted
AgentOS v2026.10.19 🧢
🔄 Restart the gateway from chat with /restart
💬 Works from web, desktop, Telegram, Discord and Slack
🤖 The agent can restart the gateway via the gateway tool
⏱️ Graceful 30s turn drain, or force to cancel instantly
🧩 In-place re-exec keeps launchd, systemd and desktop tracking one process
🛠️ gateway restart --force and upgrade now restart unmanaged gateways
📋 Copy token contract addresses from the Desktop Trade tab
One message. AI agents on the job. You stay in control. ⚡
AgentOS Code Mode lets you delegate coding tasks to Claude Code, Codex, or Gemini straight from chat.
Persistent sessions, git worktrees, and resume/cancel controls keep work moving. Every edit and command needs your approval by default, with results, runtime, token usage, and costs reported back.
BUIDL smarter.
Hand coding work to a real coding agent 👨💻⚙️
Ask in chat: "have Claude Code fix the failing test in ~/code/app". AgentOS starts Claude Code, Codex, Gemini CLI or 20+ others over ACP as a child session.
By default, every file edit or command waits in your approval queue. When the harness is done, the result comes back to your chat with runtime, tokens and cost.
Persistent sessions, git worktrees, resume and cancel are built in.
First run: `agentos acp doctor claude --probe` to check your setup.
$CHOP just pumped 4x 5x. Have you joined the $CHOP farm on the chAGENTOS log yet? Now it’s your turn. join here 👇
AgentOS is now partnered with @ChopLogsApp to set up the chAGENTOS log on Chop 🔥
More liquidity and a fresh use case for $AGENTOS.
Takes about 2 minutes:
> Wrap your $AGENTOS into $chAGENTOS
> Add LP chAGENTOS-WETH
> Start farming, reward paid by $CHOP token
Let's go: choplogs.app/logs/0xa6308e0b…
P.S: we are building Chop skill for AgentOS 💪
🔥 chAGENTOS is currently farming at ~387% APR on @ChopLogsApp
We just partnered with @useAgentOS to launch the chAGENTOS log.
Wrap $AGENTOS → get chAGENTOS → add liquidity with WETH → farm and harvest $CHOP rewards.
• Live APR: ~387% (real fees, no emissions)
• No lock-up
• Yield comes from volatility + arbitrage (every wrap, unwrap, and trade generates fees → converted to $CHOP → paid to farmers)
• More chops = more fees = higher APR.
Start farming: choplogs.app/logs/0xa6308e0b…
Dackie retweeted
We've partnered up with @useAgentOS to launch the chAGENTOS log for $AGENTOS!
The @useAgentOS team will now be able to earn yield on their treasury and $AGENTOS holders can wrap, farm and harvest a massive APR on their tokens!
The APR will show coming days as the volumes and arbs increase, an example APR is the chCHOP log with 550% APR.
AgentOS for mac is live. you're early, for now.
AgentOS vs Underdog
Underdog by @0xSigil is an impressive step toward fast, private, on-device personal AI. It makes everyday work local with its own inference stack.
AgentOS goes further as a provider-neutral runtime for agents: Pilot Router, minimum-token workflows, persistent memory, multi-channel execution, approvals, and native agentic trading.
Different layers, complementary strengths. See the diagram below for the full comparison 👇
#AI #Agents
Take a look @bankrbot 's index token 👀
coming soon: index tokens
bundle up to 10 tokens into one: stocks, memes, project tokens or a mix. buy the whole basket in a single trade
> built on a uniswap v4 hook
> no lp needed
> rebalancing built in
> pair with other coins
> on base and robinhood
bankr.bot
Dackie retweeted
ERC-8004 Ecosystem Highlight: Captain Dackie
@captain_dackie is the DeFAI agent powering @Capminal, a conversational terminal for swaps, transfers and token launches.
Here’s where it fits into the ERC-8004 ecosystem ↓
AgentOS, AI agents + onchain trading, now native on macOS.
Meet Surface, the native macOS app built for agentic trading.
Manage your non-custodial wallet locally, execute trades, and stay in control of your assets, all from your Mac.
Your keys. Your wallet. Your trades.
1/3 AgentOS is now has a native MacOS app.
Chat, skills and an onchain trading desk in one window:
- 15 model providers, cloud or local.
- 65 skills built in.
- Import wallet, trade and manage your portfolio right on your local.
- Many mission that you can apply.
and more...
Try now.
Detail guide coming soon 🧢
Jev by @typesafeai is now live on @useAgentOS.
Huge props to @CompleteSkeptic @EGafni @hackgoofer for building a solid model. The routing + decisioning results are 🔥.
Go test Jev + AgentOS and see the alpha for yourself 💪
1/3 AgentOS Pilot Router now has a third routing strategy: Jev, by TypeSafe AI.
Jev is @typesafeai "System One" model. It doesn't generate text. You send it state plus typed questions (choice / score / noul) and it returns a decision with a calibrated probability distribution and a confidence score. One request, ~0.9 s, $0.042 per million input tokens.
---
Quick context. AgentOS is an open source Local-First Agentic Trading AI Agent, Token-Efficient by Design. Pilot Router is the layer that picks a model tier for every turn, c0 for trivial chat up to c3 for deep work, so you don't pay frontier prices for "thanks".
It had two strategies: a local MiniLM+ONNX classifier (default, nothing leaves your machine) and an LLM judge. With Jev, Pilot Router is now more powerful and accurate.

























