skills.rest
toongri
toongriCOMMUNITY
Explore all skills by this author →
GitHub

agent-council

Orchestrate multi-agent council jobs for architecture and risk decisions.

25|1|Updated Jan 14, 2026
Search Tags:#multi-agent#trade-offs#claude-code#decision-support#job-orchestration#context-sync#agent-council
One-click install
npx skills add https://github.com/toongri/oh-my-toong-playground --skill agent-council-toongri
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-council
Source: https://github.com/toongri/oh-my-toong-playground/tree/main/skills/agent-council
Command: npx skills add https://github.com/toongri/oh-my-toong-playground --skill agent-council-toongri

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yaml, and includes scripts (resource) components.

What problem does it solve?

Ambiguous architecture trade-offs, subjective quality judgments, and conflicting risk assessments stall progress because no single agent can capture every perspective; this Skill solves that by running a Claude Code council that gathers multiple independent opinions so the caller can make an informed final decision.

Core Features & Use Cases

  • Multi-agent council workflow: Use the provided scripts to start the council job, collect member outputs, and clean up while respecting the foreground-only, no-sleep, and timeout (180000ms) constraints that keep deterministic execution reliable.
  • Context synchronization guardrails: Feed evaluation criteria, project context, target content, and the specific question so each member evaluates the same information before weighing architecture decisions, subjective code quality, or risk scenarios.
  • Synthesis & cleanup discipline: Read every completed output, synthesize advisory consensus and divergence using the Model Characteristics weighting table, and run the cleanup command so nothing lingers after the decision.

Quick Start

Run bun .claude/skills/agent-council/scripts/job.ts start from the project root while supplying evaluation criteria, project context, target content, and your English question to begin the council.

Frequently Asked Questions about agent-council

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

FAQPage Schema
How do I get multiple AI perspectives on architecture trade-offs in Claude Code?▼

You can gather multiple AI perspectives on architecture trade-offs by running a multi-agent council job within Claude Code, which collects independent opinions on subjective engineering judgments before you make a final decision. It orchestrates deterministic jobs with strict context synchronization and timeout limits.

What is a multi-agent council for decision support?▼

A multi-agent council for decision support is a Claude Code workflow where multiple independent AI agents evaluate the same project context and target content to provide diverse viewpoints on uncertain trade-offs. This Skill orchestrates the job, synthesizes advisory consensus and divergence, and cleans up the session.

How to start a council job for risk assessment using bun?▼

You start a council job for risk assessment by running `bun .claude/skills/agent-council/scripts/job.ts start` from the project root while supplying evaluation criteria, project context, target content, and your specific English question to begin the evaluation.

Can I use this multi-agent workflow for subjective code quality evaluation?▼

Yes, you can use this multi-agent workflow for subjective code quality evaluation by feeding evaluation criteria, project context, and target content to the council members. Each member evaluates the same synchronized information to weigh ambiguous quality judgments.

What are the execution constraints for Claude Code agent orchestration?▼

Execution constraints for Claude Code agent orchestration include a foreground-only mode, no-sleep restrictions, and a strict 180000ms timeout limit to keep deterministic execution reliable. A cleanup command must also be run after synthesizing outputs so nothing lingers.

Related Skills

Explore More Software Engineering Skills→
southlab-aisouthlab-ai
2

council-status

Display Council status with recent decisions, memory status, feed activity, and statistics.

Official
Basic
tmalcolm-0607tmalcolm-0607

council-join

Joins existing council channels with automated alias conflict resolution and atomic member list updates.

Community
Advanced
antonyfmunozantonyfmunoz

claude_agent_sdk

Orchestrate Claude Code subprocesses with budget limits and session management.

Community
Advanced
dioerdendioerden

prose

Orchestrate multi-agent AI workflows using a declarative language.

Community
Advanced