skills.rest
wopal-cn
wopal-cnOFFICIAL
Explore all skills by this author →
GitHub

agents-collab

Governs delegation to sub-agents via wopal_task tools, task lifecycle states, and notification handling.

3|1|Updated Feb 26, 2026
Search Tags:#code review#multi-agent orchestration#task lifecycle#sub-agent delegation#notification handling#workspace protection#wopal_task
One-click install
npx skills add https://github.com/wopal-cn/wopal-space-ontology --skill agents-collab-wopal-cn
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agents-collab
Source: https://github.com/wopal-cn/wopal-space-ontology/tree/main/skills/agents-collab
Command: npx skills add https://github.com/wopal-cn/wopal-space-ontology --skill agents-collab-wopal-cn

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Delegating work to sub-agents like fae and rook without clear operating rules leads to lost context, zombie tasks, wasted concurrency slots, and unrecoverable parallel-workspace corruption. This Skill defines the mandatory tool-level protocol for safe, notification-driven sub-agent collaboration. ## Core Features & Use Cases - Delegation Tool API Reference: Documents wopal_task, wopal_task_output, wopal_task_reply, wopal_task_abort, and wopal_task_finish with exact parameters and state-dependent behavior. - Task Lifecycle Management: Defines the running/idle/waiting/stuck/error state model with decision trees for handling IDLE, STUCK, and ERR notifications without polling. - Parallel Workspace Protection: Provides a mandatory prompt block forbidding git stash, checkout, reset, and other commands that could destroy other agents' uncommitted changes in shared working trees. - Use Case: When you ask the agent to "let fae implement this feature" or "have rook review my commit", this Skill loads first so the delegation prompt includes precise paths, the correct review contract (commit range vs working tree), and reuse-versus-recreate rules based on context usage thresholds. ## Quick Start Ask the agent to delegate a code implementation task to fae or a code review to rook, and this Skill will govern how the task is launched, monitored, and finished.

Frequently Asked Questions about agents-collab

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

FAQPage Schema
How do I delegate a task to a sub-agent with wopal_task?▼

Call wopal_task with a short description, a precise prompt containing the task, goal, and absolute file paths, and the target agent name such as fae or rook. It returns a task_id for later monitoring and runs asynchronously without blocking the main session.

How do I handle idle, stuck, and error sub-agent task states?▼

For idle tasks, inspect text output then either finish the task or reply for rework. For stuck tasks, check the reasoning section then reply to resume or finish to clean up. Error tasks cannot be replied to; finish them and recreate with valid configuration.

What information must a rook code review delegation prompt include?▼

For uncommitted changes, provide review_type, project_path, change_scope set to working_tree, and the changed file list or diff instruction. For committed changes, provide a commit hash or commit range plus background; never give only file names, or rook loses precise diff context.

When should I reuse a sub-agent task instead of creating a new one?▼

Prefer replying to the existing task while it is alive, scope is unchanged, and context usage is healthy. Stop reusing when context exceeds 50 percent, one rework cycle already happened at 45 percent or more, the scope materially changed, or the sub-agent is drifting or looping.

Why is git stash forbidden in parallel sub-agent delegation?▼

Parallel agents share one working tree, so git stash, checkout, reset, or clean can silently destroy another agent's uncommitted changes with no way to recover them from git history. Every parallel delegation prompt must include the explicit workspace protection prohibition block.

Why did my sub-agent task complete immediately with no real output?▼

Immediate completion without actual work usually means the delegation prompt was too vague, so the sub-agent concluded prematurely. Provide detailed steps, verification commands, and explicit completion criteria in the prompt to prevent this.

Related Skills

Explore More Software Engineering Skills→
CommunityToolkitCommunityToolkit
622

iterative-retrieval

Structures sub-agent spawning with a 3-cycle protocol, WHY context, and coordinator validation.

Official
Intermediate
buYoungbuYoung
18

linear-issue-reviewer

Validate Linear sub-issue Done Criteria against code changes and comments.

Community
Advanced
lie5860lie5860
30

subagent-driven-development

Delegate independent tasks to fresh subagents with isolated context and two-stage reviews.

Community
Advanced
mrjuguymrjuguy

spawn-child-agent

Spawn independent Claude Code agent processes for isolated sub-tasks.

Community
Advanced