1. 🦞 Meta's Nat Friedman Says the Quiet Part Out Loud: Muse Owes Its Soul to OpenClaw

The week's biggest story wasn't a launch or a benchmark — it was a confession. After users noticed eerily similar workspace filenames (down to a shared SOUL.md config convention) between Meta's new consumer agent product Muse and @openclaw, speculation exploded across X. Nat Friedman, head of product at Meta Superintelligence Labs and former GitHub CEO, addressed it head-on rather than dodging it: "we built muse from scratch, but it is definitely heavily inspired as a product by openclaw. After I used openclaw in january I bought hundreds of mac minis for the MSL team and lots of us fell in love with using openclaw (and other personal agents). @steipete is a genius and his harness was pioneering from the jump. I think a lot of people were inspired by it. our goal with muse was to build something like openclaw that we could make safe and secure and easy to use and scale to billions of people." The post racked up 5,061 likes and over 1.18 million views. OpenClaw creator Peter Steinberger (@steipete) responded with characteristic grace rather than grievance: "Because this 'Meta uses OpenClaw' story is going around. They built their own agent, being inspired. Nat and his team did a great job. Kudos!" It's a genuinely remarkable moment for the community — a scrappy personal-agent harness that started as one developer's project is now, by the admission of a Meta product lead, the acknowledged blueprint for a product that's reportedly hit #1 on the US App Store and outpaced ChatGPT's early launch numbers per TechCrunch. Whatever you think about Big Tech "inspiration," this is a validation lap OpenClaw earned.

https://x.com/natfriedman/status/2102103707936768130

2. 💸 Claude Comes Home: Official Plugin Brings Pro/Max Subscriptions Back to Hermes Agent

If you've been paying for both a Claude Pro/Max subscription and separate API tokens just to run Claude models inside @NousResearch's Hermes Agent, this week's news is a wallet-saver. Teknium (@Teknium), cofounder and lead engineer at Nous Research, announced the release of an official plugin: "Welcome back to Hermes Agent, Claude — New official plugin that uses Claude SDK without the tradeoffs to enable Claude Code subscriptions to work in Hermes Agent again! Check it out and install it here." The plugin, claude-subscription-directsdk (currently experimental at v0.3.0), routes Hermes Agent turns through the official Claude Code CLI/Agent SDK path using an existing Claude Pro/Max subscription — no separate API key, no per-token billing — while still plugging cleanly into Hermes's agent loop, tools, approvals, and context compaction. The community reaction was loud: 3,509 likes, 440,738 views, 2,359 bookmarks, and 309 replies, many of which clarified the fine print — you'll need Hermes Agent v0.21.4 or later, and you should strip any ANTHROPIC_API_KEY from your environment before installing to avoid conflicts. Practically speaking, this is exactly the kind of quality-of-life fix the community has been asking for: if you already have a Claude subscription sitting there, there's no reason to double-pay to bring Claude into your Hermes workflow.

https://x.com/Teknium/status/2102093483788107792

3. 🛠️ One Sentence In, a Working Tool Out: Hermes Agent's PM Pipeline Flexes

Independent AI/crypto news account @RoundtableSpace (part of Mario Nawfal's media operation — not Hermes Agent staff, but a consistently sharp-eyed watcher of the space) posted a demo on September 18 that's worth a rewatch if you missed it. The clip shows Hermes Agent's project-manager agent capability taking a single plain-English sentence describing an idea, classifying it, writing out a full proposal, and then dispatching sub-agents to actually build and ship a working tool — all after just one human approval tap. The demo ran through five separate one-sentence prompts, each producing a distinct, functioning tool at the end of the pipeline. It's a smaller story by the numbers (62 likes, 42,364 views, 59 bookmarks, 7 reposts), but it's a strong illustration of what makes Hermes Agent's sub-agent delegation architecture compelling: the gap between "idea" and "shipped tool" keeps shrinking, and the human's job increasingly narrows down to a single yes/no gate. Worth keeping an eye on as this pattern gets refined.

https://x.com/RoundtableSpace/status/2100808281988178354

Spotted something great on X about OpenClaw or Hermes Agent? Share it — the community is the best part.