kanban-orchestrator

Decompose requests into Kanban task graphs routed to specialist profiles.

4|Updated May 18, 2026
One-click install
npx skills add https://github.com/ZardLi1115/zedclaw --skill kanban-orchestrator-zardli1115
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kanban-orchestrator
Source: https://github.com/ZardLi1115/zedclaw/tree/main/skills/devops/kanban-orchestrator
Command: npx skills add https://github.com/ZardLi1115/zedclaw --skill kanban-orchestrator-zardli1115

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you break an ambiguous or multi-lane request into a correct Kanban task graph that routes work to the right specialists instead of tempting the model to “just do it” directly.

Core Features & Use Cases

  • Profile-grounded decomposition: discovers which assignee profiles actually exist before planning, preventing silent routing failures.
  • Anti-temptation orchestration rules: enforces “decompose, don’t execute,” and requires creating Kanban tasks for every concrete work item.
  • Parallelization and dependency gating: creates independent ready cards for parallel lanes and uses parent links to ensure children only start after required inputs finish.
  • Recovery-aware operations: guides how to reclaim, reassign, or adjust profile model settings if workers get stuck or misconfigured.
  • Common orchestrator patterns: fan-out/fan-in research-to-synthesis, pipeline with gates (planner → implementer → reviewer), and human-in-the-loop blocking/unblocking.

Quick Start

Ask the orchestrator to decompose your request and route it through Kanban using the profiles available on your machine.

Frequently Asked Questions about kanban-orchestrator

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

FAQPage Schema
How do I decompose a multi-lane request into a routed Kanban task graph?

To decompose a multi-lane request into a routed Kanban task graph, the orchestrator breaks the request into concrete work items, creates Kanban cards for each, and routes them to the correct specialist profiles without executing the work directly.

What is the best way to orchestrate parallel fan-out and fan-in workflows with dependency gating?

The best way to orchestrate parallel fan-out and fan-in workflows is by creating independent ready cards for parallel lanes and using parent links to ensure child tasks only start after required inputs finish.

How does profile-grounded task decomposition prevent silent routing failures?

Profile-grounded task decomposition prevents silent routing failures by discovering which assignee profiles actually exist on your machine before planning, ensuring concrete work items route to valid specialists.

Can I use Kanban orchestration for human-in-the-loop review and iteration cycles?

Yes, you can use Kanban orchestration for human-in-the-loop review and iteration cycles by creating blocking and unblocking patterns within the task graph to pause work until manual review is complete.

Why does the orchestrator enforce a decompose, don't execute rule for task planning?

The orchestrator enforces a decompose, don't execute rule to prevent the model from prematurely executing work, ensuring it instead creates every concrete work item as a Kanban card and reports the task structure back.

How do I recover a multi-agent pipeline when workers get stuck or misconfigured?

To recover a multi-agent pipeline when workers get stuck, the orchestrator provides recovery-aware operations that guide how to reclaim, reassign, or adjust profile model settings for misconfigured workers.