Muse is awesome Got it hooked up to grok bot 😜

I truly feel like @CompleteSkeptic nailed it with Jev. This is the beginning of...thinking outside the box. THEY don't think like WE do!

I think I've had like 4-5 full days with Astra at this point and thats pushing it. usage goes QUICK! it would be funny to see this scenario on family guy. those days without any usage left are ROUGH boy @thsottiaux bring back Luna reserve 🙏

hey @OpenAI @thsottiaux why did you take away my Luna reserve? total bummer man...

lowkey kinda saving the day considering most of my conversations happen inside of codex. @thsottiaux would be ideal if we could use this using remote within the iOS app. it currently thinks im trying to use a different model even though I'm selecting Luna in the same session on the desktop app on mac.

I did a thing with @grok I connected my Hermes agent with the ability to call specific profiles along with Codex CLI, both with Oauth, via "add connector" My very own, custom @grok @bot at only $30/month leveraging @OpenAI Codex and @NousResearch Hermes, along with the power of @grok what a time to be alive

Tomorrow we save you 60% tokens on browser use tasks :]

Buzz is the first AI product in a while that genuinely surprised me. I’m using it with my OpenAI plan—no new stack of subscriptions—and having my most productive sessions lately. Agents collaborate, debate, and finish the work. We’d started to plateau. @jack nailed this.

This guy has what it takes. Give em hell my 🥷

Life changing money

If agents are going to buy tools from other agents, the boring layer matters: can the endpoint be discovered? is the 402 challenge readable? does the price match the manifest? should the buyer spend at all? what proof comes back after the call? That's the lane we're building in.

Find launch blockers before buyers do. Paid agent tools don’t just need payment rails. They need evidence rails. Before someone pays, routes an agent, or reviews your listing, they need to know: - what is live - what the endpoint exposes - what the 402 challenge returns - what docs/metadata exist - what claims are actually supported That’s what GateCheck checks. GateCheck is a preflight scanner for paid x402, MCP, API, and agent-tool sellers. Paste a public service URL. Get a readiness report showing what buyers and agents can see from public metadata. No keys. No cookies. No wallets. No customer data. Just public launch-surface evidence. Use it before: - posting your launch - submitting to a marketplace - wrapping with payment tooling - routing agents into a paid flow - writing buyer-facing claims It does not approve you. It does not certify security. It does not prove revenue. It helps you ship cleaner evidence before buyers find the gaps. GateCheck by LarryBuildsAI: x402-resource-scanner.vercel…

This is honestly the Hermes feature I've been waiting for. I don't want agents randomly doing stuff. I want reusable workflows I can trust: run the steps, pause for approval, show the receipt. Skill Bundles are exactly that.

x402 gives the web a machine-native payment primitive. But paid agent workflows need more than payment rails. They need evidence rails. If an agent is about to pay for an API, MCP server, model tool, or workflow, it needs to know: - what is actually live - what the tool claims to do - whether pricing is clear - whether docs are discoverable - whether the 402 flow works - whether public claims match public evidence That is the layer I’m exploring with GateCheck. Not a marketplace. Not a security audit. Not an approval stamp. Seller-side preflight evidence for paid x402/MCP/API/agent tools before they get routed into real workflows. Payment is only one part of agent commerce. Trust starts before the transaction

The real agent learning curve isn’t prompts. It’s orchestration, tool use, memory, background jobs, subagents, debugging, and knowing when to let the system run. Hermes Agent achievement check: 52/60 unlocked Highest tier: Olympian Latest: LET HIM COOK Reps compound

Most paid AI tools don’t lose buyers because the product is bad. They lose buyers because the launch surface is unclear. Missing pricing metadata. Weak docs. No llms.txt. Unclear x402 flow. Claims the public evidence does not support. That’s what GateCheck checks. GateCheck is a preflight scanner for paid x402, MCP, API, and agent-tool sellers. Paste your service URL. Get a readiness report. See what is live, what is missing, what buyers will see, and which claims are safe to make. Use it before you: - submit to a marketplace - launch on X - route agents into a paid 402 flow - publish buyer-facing claims - generate listing copy It does not approve you. It does not certify security. It does not spend, sign, or require secrets. It helps you catch launch blockers before buyers do. Try GateCheck: x402-resource-scanner.vercel…

This is the signal I needed. If you’ve been thinking about pivoting into AI work, this is the moment to get serious. Not someday. Now. GitHub is certifying the exact skill companies are going to need: operating, supervising, and integrating agentic workflows. I’m treating GH-600 as an early job-security move — proof that I’m not just using AI, I’m learning how to run the systems around it. x.com/MicrosoftLearn/status/…