69 comments

[ 0.24 ms ] story [ 18.1 ms ] thread
>We take contributions as human-written text, not code — see CONTRIBUTING.md. Describe the change you'd like informally in a .txt or .md file in adrs/, and if we're aligned we'll handle the implementation. Report vulnerabilities privately — see SECURITY.md, not a public issue.

Starting to think people were right when they talked about our industry itself having an AI psychosis problem.

i find something a bit funny in an ai project, written by ai, requiring human-written text with specific guidance to not use ai.

"Given that coding agents write most underlying code now, we'd prefer PRs in the form of human-written text. [...] Please do not have AI artificially expand what you'd like to do into a formal proposal."

Is Hermes the best openclaw like agent as they mention running it before?

Also, what are power uses really using openclaw like systems for?

I think Hermes is a kitchen sink of antipatterns and bloat, which is true of the vast majority of these "Claws"
I didn't have good experience with Hermes. Vellum.ai was better, but unfortunately it had a bug where opencode go providers failed to work for a time, so I instead starting writing my own.

(Now it is working)

Hey, Marina here from vellum! What was the bug?
it says "Each deployment runs in the operator's own cloud account"

But feels like its written to run on one mac/vm and carries same drawbacks of other similar platforms. I'd rather use Hermes/Openclaw for oss or closed managed agents like Tasklet or Prajvis

Sweet, now YC itself will be the startup :D.
looks like an internal tool that yc rushed out the door to minimize any most lost ground to Buzz. That said, I'd be curious what folks think comparing these two tools.
Love seeing this direction along with Buzz.

The hardest problem in multiplayer agents, at least for us, has not been the agent loop. It is scoping and QM's per-person scopes plus shared rooms is a sane answer for a company-wide assistant.

I build in the adjacent lane, AQ (aq.dev), a multiplayer coding harness where teams run Claude Code and Codex together), so seeing YC ship "a multiplayer agent harness for work" is validating and a little surreal.

It's fascinating to see new UI primitives and concepts get invented in the LLM era. The sea of creativity makes it hard to even understand most of what each new app does, and nobody describes them well. When I went to the Hermes agent web page, I was left with zero clue about what it did or what it could do. It took a bit of digging to find the right part of the qm page that helped me grok what was going on.

I've become attached to Orca (yc-backed) for managing coding sessions in the past week, but some sort of postgres session db is what's really lacking. So, maybe it's time to try qm.

> We take contributions as human-written text, not code — see CONTRIBUTING.md. Describe the change you'd like informally in a .txt or .md file in adrs/, and if we're aligned we'll handle the implementation.

Interesting approach to open source contributions. Closer to feature requests at that point?

Interesting to see they shipped an "anti-slop" taste skill:

> description: Anti-slop frontend skill for landing pages, portfolios, and redesigns. The agent reads the brief, infers the right design direction, and ships interfaces that do not look templated. Real design systems when applicable, audit-first on redesigns, strict pre-flight check.

> - *PREMIUM-CONSUMER PALETTE BAN (mandatory, second-most-recurring AI-tell):* - For premium-consumer briefs (cookware, wellness, artisan, luxury, heritage craft, DTC home goods, etc.)... - Backgrounds: `#f5f1ea`, `#f7f5f1`...

> Landing pages and portfolios are *visual products*. Text-only pages with fake-screenshot divs are slop.

https://github.com/yc-software/qm/blob/7f2c916360f1797a8ff2a...

All due respect to the YC folks, but having a skill with 22,069 tokens is a major skill issue. Yes, ironic.

My slop control skill is a thousand tokens. Biggest problem I see with the skill is that everything is prompted via negativa. smh. sorry to be judgmental but it's hard to trust a harness that comes shipped with a skill like this.

They all use the same pill glowy status light header eyebrow to immediately tell you they have no actual design skills and follow the herd.

Posers took over the industry.

lol this is hilarious - the “make no mistakes” of design

You can’t prompt an agent to have taste.

Other than the hosting providers, who has made money directly from running OpenClaw in constant loops?

This software appears to be yet another solution in search of a problem designed to burn as many tokens as possible.

> Learn your writing voice from past sends, then triage your inbox on a schedule — labels and reply drafts included

Oh grand. The world will be full of agents talking to agents and nothing getting done. So pretty much the same only I can just leave my machine awake to look busy.

If I had applied to yc, I would have thought they had "distilled" my startup idea xD
not a very helpful title? maybe: "qm - a multiplayer agent harness for work"
I'll need to explore how they're doing org wide context and security for sure. This seems extremely complementary to my own coding tool which currently gives the best AI interface for individuals.

Would be cool if in a few epic tickets i'll have both their org wide architecture AND a productive individual coding interface :)

Just want to thank the authors for the concise and consumable README.
Aren't there a ton of products already doing this? Why not just use claude Cowork? Surely they're simpler/better/more featureful/developed than the alternatives here? What advantage does this have? Would love to see a 'QM vs Cowork' comparison!
> Would love to see a 'QM vs Cowork' comparison!

It feels like the big thing they are touting here is the shared company brain. Not clear to me though, how that brain is developed when each person has their own harness. (I did only skim the docs.)

And it still has the issue of: if the agent is acting as me, then security wise it can do anything I can do. Maybe that's why they are recommending for startups.

But yes, comparison would be helpful.

People are building state-of-the-art harnesses in private right now that outperform the average developers' harnesses, people using Claude Code, OpenCode, etc.
I gave an agent its own Slack channel and it started scheduling meetings with other agents without me. I've never felt more like middle management
I hope the other agents replied with 'This meeting could have been an email.'
In a tangential earlier release, Gary Tan's own gstack:

https://github.com/garrytan/gstack

Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA

This is the exact same as the breathless faddishness of celebrity news. "Use these seventeen different skincare products and you'll look exactly like Gwyneth Paltrow".
Oh sweet, nice to see someone writing about multivalue databases.
Cool. This also feels a lot like what copilot is doing. It’s nice that copilot integrates out of the box with teams/outlook/office, so it has great company/work context
"multiplayer"? Is this a game? I honestly don't know what that means in this context.