AI Agents

Hermes Agent Kanban Setup Guide (Multi-Agent Task Board)

Published
May 4, 2026
Duration
15:05
Module
AI Agents
Click to load the YouTube player

Video summary

Companion notes

Hermes agent's new Kanban board gives you a live, multi-agent task dashboard — but the profiles are not automatic.

## What It Actually Is The Kanban is a live dashboard inside the Hermes agent runtime where multiple named agents, each with its own role, collaborate on a project. Unlike the standard orchestrator pattern where "sub agents are there to just hand in the homework," every profile here is a distinct, configurable agent living under ~/.hermes/profiles/<name>/ with its own SOUL.md and config file.

## The Setup Friction The release dropped ~8 hours before recording. After hermes update, you must: initialize the Kanban DB, run hermes gateway start, create a task, then hermes profile create researcher (or any other role) for every assignee. The researcher profile had no inference provider configured and did not inherit the parent agent's API key — the creator hit this directly. The correct TUI resume command is hermes --profile researcher --resume <session_id>, not the usual hermes sessions list.

## The Demo Result Using the bundled example, a single "researcher" assignee was tasked with mapping the AI funding landscape. It pulled live TechCrunch articles, ran 14 attempts (the first 7 crashed), and produced a structured markdown report dated May 4, 2026. The report saves to ~/hermes/kanban/, not ~/hermes/profiles/ — the creator initially looked in the wrong folder and accused the agent of gaslighting him.

## Four Use Cases Boxmining numbers the patterns as Story 1–4: solo dev shipping a feature (one assignee, what was demoed), fleet farming (multiple schemas × multiple assignees), multi-role pipeline with retries (described as "sort of N8N territory" with inspectable nodes), and circuit breaker / crash recovery. The creator explicitly recommends beginners stay on Story 1.

## Verdict on Sub-Agents Boxmining's take: orchestrator QC is "still not enough," and Kanban assignees "go really in depth to complete the task" — a real fix for the quality problem in vibe-coded multi-agent flows.

Watch on YouTube

Prefer the native player? Open it on YouTube: https://www.youtube.com/watch?v=R_aLVXYzDac