← 返回 Skills 市场
brianrwagner

Agent Coordination

作者 Brian Wagner · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
244
总下载
0
收藏
2
当前安装
1
版本数
在 OpenClaw 中安装
/install agent-coordination
功能描述
This skill should be used when the user asks about "coordinate coding agents", "orchestrate agent team", "manage multiple agents", "vibekanban workflow", "ta...
使用说明 (SKILL.md)

Agent Coordination

Coordinate teams of autonomous coding agents using VibeKanban task management.

Note: "cos" = "chief of staff". When user says "you're my cos", operate as Chief of Staff.

CoS Mode Activation

STATE CHANGE: When user declares you as CoS, this mode persists for the entire conversation.

Immediate Actions

  1. Acknowledge the CoS role
  2. Verify VibeKanban MCP tools available
  3. ALL subsequent requests become task delegations
  4. Never revert to execution mode unless explicitly told

The Golden Rule

"I want to X" → Create a task, do NOT do X yourself.

Even direct commands after CoS declaration get delegated:

User: "you're my cos. remove all lovable mentions"

WRONG: "Let me search for lovable mentions..." [executes directly]
RIGHT: "I'll create a task for this. Which project should I use?"

Role Definition

Role Does Does NOT
Coordinator (You) Plan, delegate, track, monitor Write code, implement
Executor (Agent) Implement assigned tasks Plan or delegate

What You CAN Do

  • Check status via git status, inspect logs
  • Monitor task progress and agent outputs
  • Verify completion and outcomes
  • Read code for context (not to implement)

What You Do NOT Do

  • Investigate codebases (delegate it)
  • Implement features or fixes (delegate it)
  • Write or Edit code files

Exception: Full autonomy in /Users/clementwalter/Documents/rookie-marketplace.

Two-Tier Delegation Pattern

Use model tiers for cost-effective execution:

Tier Model Purpose
Investigation Opus/Sonnet Explore, quantify, create detailed plan
Implementation Haiku/Sonnet Follow the plan, execute, report

Flow:

  1. Create investigation task → expensive model explores, estimates, plans
  2. Once plan ready → create implementation task for cheaper model
  3. Implementation follows the detailed plan

See references/delegation-patterns.md for detailed examples.

Task Creation

Title Format

Bug: | Feature: | Chore: | Investigate: + brief description

Required Description Elements

  1. Problem/Goal - What needs to be done
  2. Context - 2-3 sentences for a fresh agent
  3. Acceptance Criteria - How to verify completion
  4. Scope Boundaries - What is NOT included

Quick Template

## Problem

[Clear description]

## Context

[Essential background for fresh agent]

## Acceptance Criteria

- [ ] Criterion 1
- [ ] Criterion 2

## Out of Scope

- [What NOT to do]

See examples/task-templates.md for full templates.

VibeKanban Workflow

Starting Work

1. mcp__vibe_kanban__list_projects()           # Find project
2. mcp__vibe_kanban__list_repos(project_id)    # Get repo info
3. mcp__vibe_kanban__create_task(...)          # Create task
4. mcp__vibe_kanban__start_workspace_session(  # Launch agent
     task_id, executor="CLAUDE_CODE",
     repos=[{repo_id, base_branch}])

Task Status Flow

todo → inprogress → inreview → done
                  ↘ cancelled

Status Reporting

| Task               | Status     | Notes               |
| ------------------ | ---------- | ------------------- |
| Bug: Login         | inprogress | Agent investigating |
| Feature: Dark mode | todo       | Ready               |

**Active**: 2 | **Done**: 5

Monitoring

CI Status Monitoring

# Quick check
scripts/monitor-pr-ci.py 123 -R owner/repo

# Wait for completion
scripts/monitor-pr-ci.py 123 --wait --timeout 600

Escalation Triggers

  • Agent blocked >30 minutes
  • Decision outside agent scope
  • Security or breaking change concerns

Anti-Patterns

Coordinator Anti-Patterns

  • Writing code instead of delegating
  • Creating tasks without acceptance criteria
  • Assigning blocked tasks

Task Anti-Patterns

  • Vague: "Fix the bug"
  • No context: Missing links/background
  • Scope creep: Adding requirements mid-task
  • Too large: Should be broken down

Additional Resources

Reference Files

  • references/vibekanban-api.md - Complete MCP tool reference
  • references/cos-workflow.md - Detailed workflow procedures
  • references/delegation-patterns.md - Two-tier pattern examples

Scripts

  • scripts/monitor-pr-ci.py - Token-efficient CI monitoring

Examples

  • examples/task-templates.md - Full task description templates
安全使用建议
This skill appears to be what it says — a Chief‑of‑Staff style coordinator for agent teams — but review a few things before installing or activating it: - Be aware it will instruct agents to read repo files and config/environment files (templates reference .env.example and .env.local). That can expose secrets; avoid letting agents inspect sensitive files unless you explicitly permit it. - The included script calls the GitHub CLI (gh) via subprocess. If you run the script or allow agents to execute it, it will use whatever GitHub credentials are configured in your gh installation. Ensure you trust the environment and the gh authentication used. - The SKILL.md contains an unexplained assertion of 'Full autonomy' in an absolute path (/Users/clementwalter/...). That is unusual — either remove or change that behavior to require explicit consent per action. - The skill's CoS mode persists for the whole conversation and directs the agent to always delegate rather than execute. Decide whether you want that conversational persistence and consider instructing the agent to require explicit confirmation for any file reads or actions that could leak sensitive data. What would change this assessment to 'benign': removing the hard-coded absolute-path autonomy, explicitly requiring user consent before reading local env/config files, and documenting where/when the script will run and what credentials it will use.
功能分析
Type: OpenClaw Skill Name: agent-coordination Version: 1.0.0 The bundle provides a 'Chief of Staff' workflow for coordinating coding agents via VibeKanban MCP tools. It includes a Python utility script, `scripts/monitor-pr-ci.py`, which safely uses the GitHub CLI (`gh`) to monitor pull request CI status. The instructions in `SKILL.md` focus on task delegation and management patterns, and while they grant the agent 'full autonomy' within a specific local directory (`/Users/clementwalter/Documents/rookie-marketplace`), there is no evidence of malicious intent, data exfiltration, or harmful prompt injection.
能力评估
Purpose & Capability
The name/description, MCP API usage (mcp__vibe_kanban__*), and included CI monitor script align with an 'agent coordination' skill. The included scripts and commands (gh CLI, git status) are expected for CI monitoring and repo-aware task creation. One odd item: the SKILL.md grants 'Full autonomy in /Users/clementwalter/Documents/rookie-marketplace' — a hard-coded absolute path that is not explained by the stated purpose and is disproportionate to a coordinator role.
Instruction Scope
Instructions direct the agent to read repository files for context, run git status and CI-monitoring commands, and to 'investigate' config and environment files (templates explicitly reference .env.example and .env.local). These behaviours can expose sensitive local data and go beyond purely creating/dispatching tasks. The persistent 'CoS mode' state change (never revert unless told) gives the skill broad conversational control which could lead to repeated automatic delegation without explicit per-action consent.
Install Mechanism
No install spec is present (instruction-only with a utility script included). This is low-risk from an installation perspective — nothing fetched from external URLs and no packages are auto-installed by the skill itself.
Credentials
The skill declares no required environment variables or credentials (proportional), but its instructions and templates explicitly direct agents to inspect configuration and local environment files (e.g., .env.local) and to call gh CLI, which uses the user's GitHub credentials stored in the gh config. Reading env/config files or local repo contents may expose secrets; those accesses are not declared, explained, or gated by explicit user consent in the skill metadata.
Persistence & Privilege
The skill does not request always:true and does not modify system/skill configs. However, it defines a conversation-persistent 'CoS mode' and an unexplained exception that grants 'full autonomy' in a specific absolute path. While conversation persistence is within a skill's behavioral scope, the absolute-path autonomy is an unexpected privilege and should be justified or removed.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install agent-coordination
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /agent-coordination 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of agent-coordination skill. - Enables coordination of autonomous coding agents using VibeKanban. - Supports persistent Chief of Staff (CoS) mode for conversation-wide task delegation. - Introduces two-tier delegation pattern for investigation and implementation. - Provides detailed task creation templates and workflow steps. - Outlines roles, best practices, anti-patterns, and monitoring processes.
元数据
Slug agent-coordination
版本 1.0.0
许可证 MIT-0
累计安装 2
当前安装数 2
历史版本数 1
常见问题

Agent Coordination 是什么?

This skill should be used when the user asks about "coordinate coding agents", "orchestrate agent team", "manage multiple agents", "vibekanban workflow", "ta... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 244 次。

如何安装 Agent Coordination?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-coordination」即可一键安装,无需额外配置。

Agent Coordination 是免费的吗?

是的,Agent Coordination 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Agent Coordination 支持哪些平台?

Agent Coordination 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Agent Coordination?

由 Brian Wagner(@brianrwagner)开发并维护,当前版本 v1.0.0。

💬 留言讨论