Abhey

Hey everyone, If you are looking for building something, but aren't sure of what to build, I made an AI agent that scans the internet for real user frustrations and uncovers potential market gaps - complete with source links. Already have something in mind? There’s a comprehensive validation flow to help you test if it’s worth building. check it out : cobuildr.tech

Hermes Agent just dropped a massive update. Big unlocks: • profiles for isolated multi-instance agents (finally!) • MCP server mode for Claude Desktop / Cursor / VS Code • official Docker support • fallback provider chains • new messaging platforms: Feishu/Lark + WeCom • Telegram webhook mode + Slack multi-workspace OAuth github.com/NousResearch/h…

The Hermes Agent update you've been waiting for is here.

@VadimStrizheus It’s like saying windows introducing a new AI feature killed ubuntu

They’ve killed OpenClaw. once again…

@zechdz @davepl1968 I literally had to do zero setup, only the one time authorisations of the corresponding skills had to be done (ofc)

@_abheygupta @davepl1968 How much setup is required. I gave it a go super vanilla just to see if it could modify itself as we went and it's turned into mush. Stopped making tool calls and just sucks ass.

OK, I installed OpenCLAW. I set it up with a backend AI server on a Dell 7875 workstation with dual Blackwell RTX6000 GPUs running DeepSeek R1 32B. And it can fall down to Qwen 2.5 for easy stuff. I created an agent to scour eBay for things I might be interested in, compile a summary, and mail it to myself. But I could have done that in shell script... so... what now? What are folks doing with it that's interesting? What are you having your agents do?

@thekitze Your own interpretability

if u are using agents (openclaw, hermes, whatever the fuck) what is the difference between using obsidian/notion dbs vs just making sqlite/postgres dbs for anything you need

@keter_slater Sounds exciting. Will give it a shot

100% this. and the config layer is where everything breaks down especially when ur stack evolves. built something called caliber that auto generates + keeps ur CLAUDE.md cursor rules etc in sync with ur actual codebase so ur agents always have accurate context github.com/caliber-ai-org…

The hard part for coding agents is increasingly not writing code. It’s everything around the code: accounts API keys billing service provisioning config permissions If more of that stack becomes agent-native from the CLI, “build and ship this end to end” starts feeling a lot more real. The moat is moving up from model quality to execution surface.

@vincentlejudge gpt5.4 Will be posting a detailed review on exactly what is better even when the models are same

@_abheygupta What model are you using for Hermes agent?

Hermes agent is so much better

@kenyondigital @nickvasiles Kinda yes. If purely managing through ai, then obsidian specifically doesn’t matter. But if it’s your second brain, and assuming you also manage it, then auto backlinks, plugins, etc help you. It’s just like a tooling layer on top of markdown

@_abheygupta @nickvasiles Yeah I hear you on the links and metadata. But I already do that in plain .md files with backlinks, frontmatter, consistent structure. Claude Code reads them natively. What's the specific thing Obsidian adds on top that I'd be missing? Just a more human-readable UI?

Whoever is taking the time to craft an insanely organized and well-documented Obsidian vault will experience personal AGI faster than anyone else. Arguably months before.

@kenyondigital @nickvasiles Markdown files are just files and have no built in relationships or context. obsidian makes them connected and structured (links nd metadata), so they’re actually usable for AI workflows.

@nickvasiles Help me understand the advantage of Obsidian over a series of markdown files and folders?

@VadimStrizheus I’m an openclaw user. Please tell me exactly what is better ?

spent the last few hours setting up Hermes agent HOLY CRAP this is way better than OpenClaw

Request for testing Give this to your openclaw instance: "update yourself to the dev channel `openclaw update --channel dev` and restart yourself. if that doesn't work -> clone github openclaw/openclaw to this machine if it's not already. then rebuild and restart yourself on main branch there" Then give your openclaw a try with your regular workflows/tasks Huge openclaw release incoming tonight, hopefully (no promises). We need to make sure we break as little as possible Plugins might break, because the plugin SDK is being refactored. Plugins will have to be refactored to use the new SDK, please do not report those Do report: native openclaw functionality that stops working Please reply under this post, we'll be checking here 👇

@travisirby @onusoz Thank you!

according to my claw - Notable fixes & improvements: Docker: Added OPENCLAW_TZ timezone support Android: Redesigned chat settings UI, fixed QR scanner onboarding iOS: Added onboarding welcome pager Windows: Suppressed visible console windows during restart/cleanup Slack: Opt-in interactive reply directives (you can now use button callbacks in Slack) Cron: Fixed isolated cron nested lane deadlocks Ollama: Hid native reasoning-only output Session: Preserved lastAccountId and lastThreadId on session reset Memory: Avoided injecting memory file twice on case-insensitive mounts Security: Prevented gateway token leak in Docker build context Codex: Polished sidebar status, agent skills, and chat rendering

@travisirby @onusoz What’s new ?

@onusoz has been working excellently for me! no issues. about 20 crons of various use cases, browser search slack and imessage

@4shadowed @heyshrutimishra Please fix the arm64 auth.scope = readonly issue 😖

@heyshrutimishra :)

OpenClaw has been quiet for too long. Something big they are cooking.

@Yampeleg Can you share its SOUL.md?😀

OpenClaw’s magic in a single screenshot:

@agentmail 👀

Last to like gets a free year of AgentMail. Good luck :)

@KSimback My agent is very straight and expressionless 🥲. Any suggestions on what to mention in SOUL.md?

For anyone who's using OpenClaw for heavy work and using lots of tokens for crons, workflows, research, etc The no-brainer move is to get the OpenAI $200/mo Pro sub and make that your base model + 5.4 coming soon Can always use Opus/Sonnet for specific tasks as needed