About
ChatGPT Dots vs GrokBot vs Hermes Agent: A Head-to-Head on Real Jobs
Jack Roberts puts ChatGPT's new personal assistant Dots head-to-head against GrokBot and Hermes Agent with four real jobs — generating an image and video through the Higgsfield API with almost no instructions, editing a podcast episode, drafting an email via Zapier, and building a competitor strategy — then breaks down cost, learning, control, and limitations. The through-line is a split by user type: Dots and GrokBot are the managed, cloud-hosted, model-locked options (OpenAI's and xAI's models respectively), while Hermes Agent is the self-hosted, model-agnostic option you install on your own hardware, pick any model for, and run in Telegram or the terminal for free. His verdict is segmented: beginners should start with GrokBot at $20 or Dots if they already pay for ChatGPT, while tinkerers who want full control and any model should run Hermes. Dots is 'a couple of weeks behind GrokBot' in maturation, but he expects OpenAI's capabilities to grow fast.
Hermes Agent Skills That Make It 10x More Powerful
Six community skills that make Hermes Agent dramatically more capable: Planning with Files (moves the plan into task_plan.md/findings.md/progress.md + a hook that forces them into context so the agent can't drift), Delegate (turns Hermes into an orchestrator across Claude Code/Codex/Cursor), RTK (filters terminal output so 90% of noise never hits the context window), Mantis (Google's security review skills, scheduled via Hermes cron jobs), Agent Reach (free access to 15+ restricted platforms like Reddit), and Skill Retriever (matches your message against your skill list, cutting the index from ~11K to ~2.3K tokens per turn).
Hermes in Buzz: Three Agents on Three Machines, One Shared Workspace
A fifteen-minute setup walkthrough: three Hermes agents on three different machines (PC, DGX Spark, VPS) communicating through a Buzz relay. The setup uses the native Hermes Gateway (recommended over the desktop ACP), requires the Buzz CLI binary on path — the single gotcha that produces a misleading channel-membership error when the CLI is missing — and is assisted by a dedicated skill from the presenter's buzz-skills repo (159 stars). The closing demo shows two Hermes agents cross-checking research: one researches Buzz updates and reports them, the other verifies and catches an error — a merge specification misidentified as an implementation change. A Claude agent also runs in the same workspace.
Buzz: A Shared Room for Agents, and the Trust Problem Underneath
Block's Buzz took 17,010 GitHub stars in under five months by putting Claude Code, Codex and Grok in the same channel, driving the real CLI tools through the open Agent Client Protocol. The 'free Slack clone' framing undersells a self-hostable Nostr relay where every message and git event is a signed, attributable entry in one log. The most valuable stretch is where it fails: a 9am scheduled task that announced what it was about to do and then did not do it, and a presenter admitting he does not trust his own agent team.
14 Hermes Agent Skills Worth Installing — Tested Against Hundreds
Dubi ships apps daily with Hermes Agent and has worked through hundreds of community skills. This is the shortlist that survived, ranked 14 to 1: SkillClaw's self-evolving library, Defuddle's token savings, oh-my-hermes multi-agent verification, and Agent-Reach at number one for giving agents platform access they simply did not have. Plus the 15KB rule that separates a good skill from bloat.
The Ultimate Beginners Guide to Hermes Agent
Comprehensive 3-hour course by Samin Yasar covering full Hermes Agent setup, channels (Telegram/Slack/iMessage), memory (soul.md/user.md), Honcho shared brain, Multica project board, profiles and squads, cron jobs, and 5 full builds: morning briefing, animated website, property kit, content octopus, trading bot. Plus Jarvis command center and how to sell your builds.
Hermes Agent + Mixture of Agents is insane…
David Ondrej demonstrates Hermes Agent's Mixture of Agents (MOA) feature — orchestrating multiple AI models (GLM 5.2, GPT-5.5, Kimi K2.7, Opus 4.8) through a single aggregator to achieve frontier-level intelligence, with a full VPS setup walkthrough and live 3D game build demo.
Hermes Architecture EXPLAINED — Memory, Context & Gateways
Alejandro AO walks through the full Hermes Agent architecture — the agent loop, context construction from markdown files, context compression strategies, the gateway system for Telegram/Slack/Discord, three-tier memory (markdown + SQLite + external providers), and the cron job scheduler.
Hermes Agent Quickstart — From Zero to a Working Setup
Everything you need to install Hermes Agent, connect your preferred LLM provider, and get a working AI assistant — from a one-line install to messaging gateway, skills, voice mode, and MCP integration.
Remote Access to Hermes Agent — Deploy the Desktop App and Connect to a Remote Backend
A complete step-by-step guide to deploying Hermes Agent on a remote server and connecting to it from the Desktop App on your laptop. Covers Tailscale VPN, SSH tunnels, systemd services, and security best practices.