用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/glawson6/jaiclaw --skill kanban-orchestrator命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
This skill should be used when the user asks to "draw a diagram", "show a status box", "render ASCII art", "make a callout", "wrap text in a box", "plot these numbers", or any request involving boxed messages, two-box-and-arrow diagrams, scatter plots, or tables rendered as ASCII. Uses JBang to invoke the JaiClaw ASCII renderer; do not hand-draw borders character-by-character.
Render domain objects (events, tasks, tickets, orders, …) as framed monospaced ASCII via the render_response tool. Follow the fidelity rules exactly — do not paraphrase, transliterate, forge, or strip borders from tool output.
End-to-end validation of JaiClaw bootstrap, scaffolding, build, runtime, and Docker CLI images. Tests quickstart.sh, project creation from Maven releases, provider connectivity, CLI launcher, and Docker image builds.
基于 SOC 职业分类
正在显示 SKILL.md
| name | kanban-orchestrator |
| description | Decompose and route work through multi-agent Kanban systems |
| alwaysInclude | false |
| requiredBins | [] |
| platforms | ["darwin","linux"] |
| version | 1.0.0 |
| tenantIds | [] |
Coordinate multi-agent workflows using a Kanban-style board. The orchestrator decomposes complex tasks into discrete work items, assigns them to worker agents, and tracks progress through lifecycle stages.
Maintain a Kanban board with these columns:
| Column | Purpose |
|---|---|
| Backlog | Decomposed tasks awaiting assignment |
| In Progress | Tasks actively being worked by an agent |
| Review | Completed tasks awaiting verification |
| Done | Verified and accepted tasks |
| Blocked | Tasks with unresolved dependencies |
When receiving a complex request:
small, medium, or large.id: TASK-001
title: Short imperative title
description: |
Detailed description of the work to be done.
Include inputs, expected outputs, and constraints.
acceptance_criteria:
- Criterion 1
- Criterion 2
blocked_by: []
complexity: small | medium | large
worker_type: coding | research | testing | review
Use background processes to run workers in parallel:
# Dispatch to a coding worker
bash background:true workdir:/path/to/project command:"claude --print 'TASK-001: Implement the login form component'"
# Dispatch to a research worker
bash background:true command:"claude --print 'TASK-002: Research OAuth 2.0 PKCE flow best practices'"
Monitor all active workers:
process action:poll sessionId:WORKER_SESSION_ID
process action:log sessionId:WORKER_SESSION_ID