Junai FAQs

Question 1

How does Junai ensure deterministic and predictable AI agent routing?

Accepted Answer

Junai achieves determinism through a Python state machine that owns all routing logic, ensuring zero hallucinated transitions. Key decisions require explicit approval via 'deterministic gates,' guaranteeing predictable output across its 9 stages and 24 agents.

Question 2

How do I interact with Junai within my development environment?

Accepted Answer

Junai provides a chat-first user experience directly within GitHub Copilot Chat in VS Code. You can issue commands to initialize, reset, switch modes, and approve gates, making pipeline management seamless without leaving your coding environment.

Question 3

Can I customize the level of automation with Junai?

Accepted Answer

Yes, Junai offers three flexible pipeline modes: supervised (you approve everything), assisted (agents route, you approve gates), and autopilot (highly automated with smart gates, requiring only intent approval). You can switch between modes anytime via Copilot Chat.

Question 4

What is Junai and how does it improve developer workflows?

Accepted Answer

Junai is a portable agent framework for VS Code and GitHub Copilot that orchestrates 24 specialized AI agents through a deterministic 9-stage pipeline. It ensures predictable, auditable outcomes by preventing AI hallucination in routing, making your development process reliable and efficient.

Question 5

What kind of AI agents are available in Junai?

Accepted Answer

Junai includes 24 specialized AI agents covering various development roles, such as Architect, Implementer, Tester, Code Reviewer, Debugger, Security Analyst, and more. These agents leverage 70+ reusable skills and prompts for comprehensive task execution.

Junai

Junai

主な機能

ユースケース

主な機能

ユースケース