Can the council automatically modify my project files?

No. The skill includes mandatory implementation gates that require explicit user approval for a synthesis or plan before any code changes are executed.

What is the difference between the depth settings?

Quick provides a fast, low-cost pass; Standard offers balanced coverage; and Deep includes intensive rounds of critique and revision for high-stakes decisions.

Does this skill cost more to run than standard Claude commands?

Yes, as it calls multiple models simultaneously. However, it includes configurable budget caps and depth settings to help you manage and limit costs effectively.

What models can I use in the council?

It is runner-backed and supports various providers. You can pin specific personas or allow the system to auto-select models based on availability and quorum requirements.

What is the main benefit of the Council skill?

It uses multiple AI models to critique each other's outputs, surfacing risks and blindspots that a single model might miss during complex tasks.

Council Multi-LLM Orchestrator

The Council skill provides a robust framework for multi-agent collaboration within Claude Code, allowing users to leverage up to eight different AI models simultaneously to solve complex problems. By implementing a structured workflow—including independent persona-based advice, cross-model critique, and a final synthesis from a 'chair' model—it surfaces hidden risks and reduces hallucinations. This skill is essential for high-stakes architectural decisions, security reviews, and complex implementation planning where single-model limitations might lead to errors.

Key Features

01Automated synthesis featuring agreement reports and risk registers

023,590 GitHub stars

03Configurable execution depth (Quick, Standard, Deep) with budget caps

04Integration with project research and durable corpus storage

05Safety veto gates for security, legal, and architectural compliance

06Multi-LLM cross-critique and independent persona-based advice

Use Cases

01Evaluating complex software architectural patterns and trade-offs

02Synthesizing research findings from large project documentation corpuses

03Conducting multi-perspective security and red-team audits

Tieline — product intent grounded in code

Council Multi-LLM Orchestrator

Key Features

Use Cases

OpenClaw Channel Configurator

Sor Ijf

GH Issues Auto-Fixer

Key Features

Use Cases