|
| 1 | +# 0.2.0 |
| 2 | + |
| 3 | +> Published 5 Jun 2025 |
| 4 | +
|
| 5 | +## Features |
| 6 | +- Add media types (image/audio/document) support to prompt API and models (#195) |
| 7 | +- Add token count and timestamp support to Message.Response, add Tokenizer and MessageTokenizer feature (#184) |
| 8 | +- Add LLM capability for caching, supported in anthropic mode (#208) |
| 9 | +- Add new LLM configurations for Groq, Meta, and Alibaba (#155) |
| 10 | +- Extend OpenAIClientSettings with chat completions API path and embeddings API path to make it configurable (#182) |
| 11 | + |
| 12 | +## Improvements |
| 13 | +- Mark prompt builders with PromptDSL (#200) |
| 14 | +- Make LLM provider not sealed to allow it's extension (#204) |
| 15 | +- Ollama reworked model management API (#161) |
| 16 | +- Unify PromptExecutor and AIAgentPipeline API for LLMCall events (#186) |
| 17 | +- Update Gemini 2.5 Pro capabilities for tool support |
| 18 | +- Add dynamic model discovery and fix tool call IDs for Ollama client (#144) |
| 19 | +- Enhance the Ollama model definitions (#149) |
| 20 | +- Enhance event handlers with more available information (#212) |
| 21 | + |
| 22 | +## Bug Fixes |
| 23 | +- Fix LLM requests with disabled tools, fix prompt messages update (#192) |
| 24 | +- Fix structured output JSON descriptions missing after serialization (#191) |
| 25 | +- Fix Ollama not calling tools (#151) |
| 26 | +- Pass format and options parameters in Ollama request DTO (#153) |
| 27 | +- Support for Long, Double, List, and data classes as tool arguments for tools from callable functions (#210) |
| 28 | + |
| 29 | +## Examples |
| 30 | +- Add demo Android app to examples (#132) |
| 31 | +- Add example with media types - generating Instagram post description by images (#195) |
| 32 | + |
| 33 | +## Removals |
| 34 | +- Remove simpleChatAgent (#127) |
| 35 | + |
| 36 | +# 0.1.0 (Initial Release) |
| 37 | + |
| 38 | +> Published 21 May 2025 |
| 39 | +
|
| 40 | +The first public release of Koog, a Kotlin-based framework designed to build and run AI agents entirely in idiomatic Kotlin. |
| 41 | + |
| 42 | +## Key Features |
| 43 | + |
| 44 | +- **Pure Kotlin implementation**: Build AI agents entirely in natural and idiomatic Kotlin |
| 45 | +- **MCP integration**: Connect to Model Context Protocol for enhanced model management |
| 46 | +- **Embedding capabilities**: Use vector embeddings for semantic search and knowledge retrieval |
| 47 | +- **Custom tool creation**: Extend your agents with tools that access external systems and APIs |
| 48 | +- **Ready-to-use components**: Speed up development with pre-built solutions for common AI engineering challenges |
| 49 | +- **Intelligent history compression**: Optimize token usage while maintaining conversation context |
| 50 | +- **Powerful Streaming API**: Process responses in real-time with streaming support and parallel tool calls |
| 51 | +- **Persistent agent memory**: Enable knowledge retention across sessions and different agents |
| 52 | +- **Comprehensive tracing**: Debug and monitor agent execution with detailed tracing |
| 53 | +- **Flexible graph workflows**: Design complex agent behaviors using intuitive graph-based workflows |
| 54 | +- **Modular feature system**: Customize agent capabilities through a composable architecture |
| 55 | +- **Scalable architecture**: Handle workloads from simple chatbots to enterprise applications |
| 56 | +- **Multiplatform**: Run agents on both JVM and JS targets with Kotlin Multiplatform |
| 57 | + |
| 58 | +## Supported LLM Providers |
| 59 | + |
| 60 | +- Google |
| 61 | +- OpenAI |
| 62 | +- Anthropic |
| 63 | +- OpenRouter |
| 64 | +- Ollama |
| 65 | + |
| 66 | +## Supported Targets |
| 67 | + |
| 68 | +- JVM (requires JDK 17 or higher) |
| 69 | +- JavaScript |
0 commit comments