Night Shift v5.0 Release Notes

We are thrilled to announce Night Shift v5.0, a significant leap forward in autonomous agent capabilities! Inspired by the SuperClaude Framework, this release introduces a robust Cognitive Architecture designed to make Night Shift agents dramatically more intelligent, reliable, and token-efficient.

✨ Highlights of v5.0

🧠 Cognitive Architecture

Night Shift now operates with an advanced cognitive framework, enabling agents to think and act like specialized experts.

• Reflexion Memory: A structured learning system that stores and reuses successful error-fix patterns, preventing past mistakes from recurring.
• Context-Oriented Personas: Agents adopt specialized mindsets (e.g., backend-architect, security-engineer) based on task needs, guided by rich Markdown definitions.
• Confidence Check (Pre-flight): An intelligent pre-execution validation that assesses task feasibility and readiness, reducing token waste from "running blind."
• Self-Check Protocol (Post-flight): A rigorous post-execution verification system that enforces strict quality gates (the "4 Questions") to ensure "true completion" with evidence, tailored to each persona's expertise.

🎭 Enhanced Persona System

• 20+ Specialized Personas: Integrated a comprehensive suite of SuperClaude-inspired expert personas (e.g., system-architect, python-expert, deep-research-agent) to cover a wide range of engineering, quality, analysis, and management tasks.
• Dynamic Persona Selection: Intelligent persona_rules in settings.yaml automatically select the most appropriate expert for a given task description, ensuring specialized focus without manual intervention.

🚀 Token Efficiency & Scalability

• Layer 0 Bootstrap: A smart context loading mechanism that initially provides only the most essential project overview (file tree, README.md), dramatically reducing initial token consumption.
• Progressive Loading: Future enhancements will build upon this foundation to load additional context only as needed, further optimizing large project handling.

🛡️ Robust Quality Gates

• The new Quality Gates system, powered by Confidence Check and Self-Check Protocol, ensures that tasks are not just "completed" but executed correctly and completely, adhering to high standards and minimizing rework.

📈 What's New

Core Features

• Introduced ReflexionMemory for structured error learning.
• Implemented ContextLoader for Markdown-based persona definitions.
• Added ConfidenceChecker for pre-execution task validation.
• Integrated SelfCheckProtocol for persona-aware post-execution quality assurance.
• Developed TokenOptimizer with Layer 0 Bootstrap for efficient context loading.
• Updated orchestrator.py to integrate all new cognitive modules and version information (v5.0).

Documentation & Templates

• README.md updated to highlight v5.0 features.
• docs/features.md and docs/architecture.md comprehensively updated to reflect the new cognitive architecture and persona system.
• New docs/quality_gates.md document explaining the multi-stage completion verification process.
• All docs/templates/*.yaml mission templates updated to use new persona names and the task: field for consistency.
• settings.example.yaml updated with extended persona_rules for automatic persona selection.

🛠️ Fixes & Improvements

• Restored validate_mission_schema and validate_settings_schema functions in nightshift/validation.py to resolve import errors.
• SelfCheckProtocol now intelligently adapts its validation checks based on the assigned persona and task type (e.g., skipping tests_passed for research tasks).

Get Started

To leverage these new capabilities, simply update your Night Shift installation and review the updated README.md and docs/ folder for comprehensive guidance.

We're excited for you to experience a new level of autonomous efficiency and reliability with Night Shift v5.0!