Council Gate FAQs

Question 1

Can I integrate this into my existing GitHub Actions workflow?

Accepted Answer

Yes, Council Gate is designed specifically for CI/CD pipelines and includes a standardized GitHub Actions configuration for easy implementation.

Question 2

Does this skill support different focus areas like security?

Accepted Answer

Yes, you can configure the 'rubric-focus' parameter to specifically evaluate code for Security, Performance, or other domain-specific standards.

Question 3

What happens if the AI models cannot reach a clear consensus?

Accepted Answer

The skill returns a specific exit code (2) for 'UNCLEAR,' which allows the pipeline to pause and request a human review instead of failing or passing incorrectly.

Question 4

How does Council Gate ensure accuracy in its verdicts?

Accepted Answer

It uses a multi-model consensus mechanism, where several LLMs must agree on a verdict based on a specific rubric and confidence threshold, significantly reducing errors compared to single-model evaluations.

Question 5

Where are the audit logs for these AI decisions stored?

Accepted Answer

All deliberations—including model responses, peer reviews, and the final synthesis—are saved in the .council/logs/ directory for full auditability and transparency.

Council Gate

主要功能

使用场景

Council Gate

主要功能

使用场景