Skip to content

CHANGELOG.md

Latest commit

 

History

History
141 lines (112 loc) · 4.48 KB

CHANGELOG.md

File metadata and controls

141 lines (112 loc) · 4.48 KB

Changelog

All notable changes to the OpenCode Orchestrator Plugin will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.1.0] - 2025-11-15

Added - MAJOR: Agent-Activated Design

  • 🚨 BREAKING CHANGE: Orchestrator now only activates with specific agents
  • Added two orchestrator agents:
    • auto-optimized: Cost-efficient development mode
    • auto-performance: Performance-optimized development mode
  • Agent activation prevents interference with other plugins and workflows
  • Added activeAgents configuration option
  • Added agent detection logic in plugin hooks
  • Added AGENT-SETUP.md comprehensive guide

Added - Agent Features

  • Agent-specific strategies (cost-optimized, performance-optimized, balanced)
  • Visual feedback when orchestrator is active/inactive
  • Automatic agent detection from session info
  • Support for custom orchestrator agents

Changed

  • Plugin now checks for active agents before running orchestration
  • Updated all configuration examples with activeAgents field
  • Updated README with agent activation instructions
  • Updated QUICKSTART with agent installation steps
  • Updated install.sh to install agent configuration files

Documentation

  • New AGENT-SETUP.md with complete agent usage guide
  • Updated README with prominent agent activation section
  • Updated QUICKSTART with agent activation steps
  • Updated installation instructions to include agents
  • Added troubleshooting for agent activation

Migration Guide

If upgrading from 1.0.0:

  1. Install the agent files: cp agents/*.md ~/.config/opencode/agent/
  2. Add activeAgents to your config: 
    activeAgents:
  - auto-optimized
  - auto-performance
  3. Switch to an orchestrator agent when you want auto-selection

1.0.0 - 2025-11-15

Added

  • Initial release of OpenCode Orchestrator Plugin
  • Automatic model selection based on task complexity
  • Multi-factor complexity detection:
    • Keyword matching
    • Pattern recognition (regex)
    • Token count analysis
    • Code complexity analysis
    • File count estimation
  • Task type override system for specialized workflows:
    • Planning tasks
    • Code review tasks
    • Debugging tasks
    • Documentation tasks
    • Quick fixes
  • File pattern overrides for critical files
  • Cost optimization features:
    • Smart model downgrade when appropriate
    • Maximum cost per request limiting
    • Fallback chain for model failures
  • Comprehensive logging system (silent, minimal, normal, verbose)
  • Custom checkComplexity tool for manual analysis
  • TypeScript type definitions
  • Two example configurations:
    • Generic example for all users
    • User-optimized example (GPT-5 Codex + Claude + GLM)
  • Automated installation script
  • Full documentation:
    • Comprehensive README
    • Quick Start Guide
    • Inline code comments

Configuration Options

  • Support for global configuration (~/.config/opencode/orchestrator.config.md)
  • Support for per-project configuration (.opencode/orchestrator.config.md)
  • Markdown configuration format with YAML frontmatter
  • Alternative YAML configuration format

Detection Features

  • Four complexity levels: simple, medium, complex, advanced
  • Configurable indicators for each level
  • Weighted scoring system
  • Confidence scoring
  • Task type detection and override
  • File pattern-based overrides

Cost Optimization

  • Estimated 60-70% cost savings for typical workloads
  • Intelligent model tier selection
  • Cost per request limiting
  • Downgradable model selection

Developer Experience

  • Transparent decision-making with detailed reasoning
  • Real-time logging of model selection
  • Custom tool for complexity checking
  • Easy customization via configuration
  • Examples for common use cases

[Unreleased]

Planned Features

  • Machine learning-based complexity detection
  • A/B testing support for model comparison
  • Integration with cost tracking APIs
  • Model performance benchmarking
  • Web UI for configuration management
  • Support for model ensembles
  • Automatic configuration optimization based on usage patterns
  • Support for custom scoring algorithms
  • Integration with OpenCode analytics
  • Multi-language support for keywords
  • Context-aware model selection (based on conversation history)
  • User feedback integration for improving selection