skills.rest
alirezarezvani
alirezarezvaniCOMMUNITY
Explore all skills by this author →
Website𝕏@RezaRezvaniBlnGitHub

hub-init

Initializes an AgentHub multi-agent collaboration session with task, agent count, and evaluation criteria.

25.3k|3.6k|Updated Oct 19, 2025
Search Tags:#parallel agents#evaluation#CLI#benchmark#session initialization#AgentHub#Multi-Agent
One-click install
npx skills add https://github.com/alirezarezvani/claude-skills --skill hub-init
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: hub-init
Source: https://github.com/alirezarezvani/claude-skills/tree/main/engineering/agenthub/skills/hub-init
Command: npx skills add https://github.com/alirezarezvani/claude-skills --skill hub-init

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Setting up a multi-agent competition requires manually creating session directories, generating session IDs, and configuring evaluation criteria, which is repetitive and error-prone.

Core Features & Use Cases

  • Session Initialization: Creates the .agenthub/ directory structure with a unique session ID and stores task, agent count, and evaluation configuration.
  • Dual Evaluation Modes: Supports command-based evaluation with a named metric and direction, or LLM judge mode when no eval command is provided.
  • Baseline Capture: Runs the eval command once at setup and records a baseline metric for later delta comparison during ranking.
  • Use Case: Run /hub:hub-init --task "Optimize API" --agents 3 --eval "pytest bench.py" --metric p50_ms --direction lower to start a three-agent optimization competition with a measured baseline.

Quick Start

Ask the AI to initialize an AgentHub session for your task with a chosen number of agents and an optional evaluation command.

Frequently Asked Questions about hub-init

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

FAQPage Schema
How do I start a multi-agent competition session with AgentHub?▼

Run /hub:hub-init with a task, agent count, and optional eval command, for example --task "Optimize API" --agents 3 --eval "pytest bench.py" --metric p50_ms --direction lower. This creates the session directory and configuration, then you launch agents with /hub:spawn.

How do I initialize an AgentHub session without an eval command?▼

Omit the --eval, --metric, and --direction arguments to enter LLM judge mode. The session is created with no automated metric, and agent outputs are compared by an LLM judge instead of a benchmark command.

What happens if the baseline eval command fails during hub-init?▼

The skill warns you but continues with session creation, since the baseline is optional. The baseline value is only used later by result_ranker.py to show metric deltas, so the session remains fully usable without it.

Can I run hub-init interactively without arguments?▼

Yes, running /hub:hub-init with no arguments starts interactive mode, which prompts for the task, agent count (default 3), eval command, metric, direction, and base branch (default current branch).

What is the next step after initializing an AgentHub session?▼

After init completes, run /hub:spawn to launch the configured number of agents, or /hub:spawn {session-id} if multiple sessions exist. The init output displays the session ID and next-step instructions.

Related Skills

Explore More Software Engineering Skills→
YunosukeYoshinoYunosukeYoshino

review

Launch parallel code, performance, and specification review agents and consolidate findings.

Community
Advanced
Patasse97Patasse97

run

Orchestrate the full AgentHub lifecycle in a single command.

Community
Advanced
hbvg234hbvg234
1

agenthub

Coordinate parallel AI agents in isolated git worktrees and merge the best result.

Community
Advanced
keugenekkeugenek
12

shark-status

Check Shark background job status, completed results, and recent logs.

Community
Basic