kanban-worker

Guide Hermes Kanban workers on lifecycle, workspace, and handoff semantics.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/MarbleSodas/Mavis --skill kanban-worker-marblesodas
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: kanban-worker
Source: https://github.com/MarbleSodas/Mavis/tree/main/skills/devops/kanban-worker
Command: npx skills add https://github.com/MarbleSodas/Mavis --skill kanban-worker-marblesodas

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents Kanban workers from making brittle assumptions and missing important handoff, retry, and completion semantics in the Hermes Kanban workflow.

Core Features & Use Cases

  • Workspace correctness: Guides how to behave in scratch, shared persistent dirs, and Git worktrees so state is handled safely across runs.
  • Tenant isolation discipline: Ensures per-tenant memory and persistent entries are namespaced to prevent context leaks.
  • Completion and handoff shaping: Defines how to use kanban_complete, kanban_block (with review-required conventions), kanban_comment, and created_cards without hallucinating IDs.
  • Operational safety checks: Covers task state changes, stale artifacts, retry diagnostics, and what to do when runs are blocked or archived.
  • Heartbeat and retry guidance: Recommends useful heartbeat content and interprets common run outcomes (timed_out, crashed, spawn_failed, reclaimed, blocked).

Quick Start

Load the kanban-worker Skill when you are dispatched as a Hermes Kanban worker so you follow the correct workspace, tenant, completion, and retry behaviors for the task.

Frequently Asked Questions about kanban-worker

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent phantom task IDs when completing Kanban worker runs?▼

Kanban worker workspace state is managed by distinguishing between scratch directories and shared persistent dirs or Git worktrees. You must follow specific workspace handling guidance to ensure state is preserved safely across multiple dispatched runs.

What is the correct way to handle tenant isolation in a multi-agent Kanban workflow?▼

Tenant isolation in a multi-agent Kanban workflow requires namespacing per-tenant memory and persistent entries. This discipline prevents context leaks between tenants when multiple workers process tasks concurrently within the same Hermes environment.

How do I handle blocked, timed out, or crashed Kanban worker runs?▼

Handling blocked, timed_out, or crashed Kanban worker runs involves interpreting run outcomes through retry diagnostics and heartbeats. You should use kanban_block with review-required conventions and provide useful heartbeat content to explain the operational state.

When do I need to use kanban_block instead of kanban_complete for task handoffs?▼

You need to use kanban_block instead of kanban_complete when a task requires review gating or cannot proceed safely. kanban_block applies review-required conventions, whereas kanban_complete is only for validated, finished tasks without phantom IDs.

Does the Kanban worker lifecycle support retries with persistent workspaces?▼

Yes, the Kanban worker lifecycle supports retries with persistent workspaces by providing specific retry guidance and operational safety checks. It interprets outcomes like reclaimed or spawn_failed while ensuring stale artifacts are managed correctly across runs.