I Built an Autonomous Cross-Agent Workflow (ClaudeCode to Hermes and Back)
One agent builds, a different agent audits it, then the first agent double-checks the audit — a cross-agent workflow that runs overnight while I watch baseball. Sign up for my FREE weekly newsletter, where I spill my unfiltered thoughts on the latest AI news, cool research, and projects I'm building: https://www.onchainaigarage.com/ 🐦 Follow Tonbi on X for real-time AI x blockchain updates! https://x.com/tonbistudio Most of my multi-agent workflows have lived inside one harness — this one goes cross-agent, spanning Claude Code and Hermes Agent, and it's the real pipeline I use to create new wikis for Agent Wikis (agentwikis.com). The shape: I pick the topics, Claude (Opus 4.8) does the heavy lifting of building each wiki from scratch, Hermes (GPT-5.5) audits the result against a strict review spec, and then it hands back to Claude to reconcile — deliberately not trusting the audit, but verifying every finding so I don't "fix" false positives. The glue is a single wake-agent Cron job that costs nothing until Claude writes "finished," then triggers the audit and the handoff automatically. I walk through every piece — the build script, the review.md spec, the wake gate, the /goal reconciliation, and the manual check — then let my VPS agent Admiral sync the finished wikis live. Resources: 🔗 Agent Wikis (the live site): https://agentwikis.com/ 🔗 Hermes Agent: https://github.com/NousResearch/herme... Timestamps: 0:00 - The cross-agent audit workflow (and why) 2:28 - Pipeline overview: me to Claude to Hermes to Claude 3:33 - Step 1: picking topics from demand signals 5:07 - Step 2: Claude Code builds the wiki 7:18 - Step 3: the wake-agent gate + audit Cron job 10:46 - Hermes audits, Claude reconciles the findings 12:38 - Reading the reports: false positives vs real fixes 16:08 - Verifying the loop + my manual check 17:46 - Deploy: Admiral syncs it live to the site I know there are other ways to wire a cross-agent audit like this, maybe even cleaner ones — if you've found one that works better, drop it in the comments. If this was useful, please like, subscribe, and hit the bell! 🦐✨ #AIAgents #MultiAgent #ClaudeCode #HermesAgent #AgenticWorkflows #Automation #Cron #AITools #AgentWikis

Install GrapheneOS Before Your Phone Becomes the Checkpoint

I stopped using /grill-me for coding. Here’s what I use instead:

Building Great Agent Skills: The Missing Manual

Trying Out Matt Pocock's /grill-me Skill in Hermes Agent

I Killed Garry Tan's GBrain Before It Killed My Agent. I Gave Hermes Brain, It Got Amnesia.

Give Your AI Agent a Second Brain (Gbrain + Hermes Agent)

Hermes Agent Masterclass: 9. Profiles & Kanban

Loop Engineering Totally 10x Hermes agents

PocketBase: A Full Backend in One File (Firebase Without the Bill)

Build an OKF brain like mine!

7 Hermes uses you must try (that actually work)

How To Build A Self-Improving AI Trading Agent (Insanely Cool)

Hermes Agent Masterclass: 8. Subagents & Delegation

Build a Multi-Agent Kanban Workflow in 20 Minutes ⚚ Hermes Automation (No Coding!)

Obsidian Just 10x’d Everyone’s Hermes Agent

100 hours of Hermes Agent lessons in 19 minutes

Creating a New Fable? Testing Mixture of Agents in Hermes Agent

Everything That Actually Matters for Local AI

Claude’s New Trading Agent Is Insane! (Tutorial)

