Download Latest Version
v2.2.7 source code.zip (7.6 MB)
Get an email when there's a new version of Agents-Flex
| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-08-10 | 3.4 kB | |
| v2.2.7 source code.tar.gz | 2026-08-10 | 6.4 MB | |
| v2.2.7 source code.zip | 2026-08-10 | 7.6 MB | |
| Totals: 3 Items | 14.1 MB | 0 | |
- docs(website): add document extractors documentation and enable MPA mode
- docs: remove en documentation files and samples
- docs(community): update communication and contribution guidelines
- style(docs): refine homepage layout and content structure
- refactor(docs): update theme layout and styling for enhanced accessibility
- docs(menu): rename Tool to Function Call in navigation and links
- docs(sidebar): refactor documentation structure and remove cursor enhancement
- refactor(doc-extractor): move document extraction components to dedicated module
- feat(agent): add conditional tool group and context compression support
- feat(store): add context compression state persistence with JDBC and Redis support
- feat(agent): add context compression coordinator for incremental history management
- refactor(agent): remove compactBeforeContextCompression option
- feat(agent): add option to normalize tool turns before semantic compression
- refactor(model-router): update ChatModelSelector interface to use ChatModelCandidates
- feat(model-router): add custom selector support for RoutedChatModel
- feat(agent): add multimodal chat model support
- feat(compression): add per-message context compression strategy
- feat(agent): add incremental context compression feature
- feat(agent): add semantic summarization with chat model
- feat(agent): enhance context compression with system message support and local strategies
- feat(agent): add semantic context compression and recent turn protection
- feat(agent): add context window management with turn-based limits
- fix(router): enhance circuit breaker and retry mechanism with endpoint deduplication
- feat(embedding): add RoutedEmbeddingModel with load balancing and circuit breaking
- feat(model-router): enhance RoutedChatModel with streaming fault tolerance
- feat(chat): enhance RoutedChatModel with robust streaming and retry mechanisms
- feat(model): add structured model error classification and retry handling
- docs(agent): add execution policy and tool context documentation
- refactor(agent): move tool error message factory from agent to execution policy
- feat(agent): add ToolErrorMessageFactory for customizing tool error messages
- feat(chat): enable ToolGroup to provide chat interceptors dynamically
- refactor(core): move tool group classes to dedicated package Michael
- refactor(chat): move streaming state from ChatOptions to ChatContext
- refactor(util): rename isEmpty method to isMetadataEmpty in Metadata class
- feat(chat): implement request-level ChatOptions copying for thread safety
- refactor(chat): enhance ChatInterceptorProvider with registration support
- feat(chat): add ChatInterceptorProvider support for dynamic request-level interceptors
- refactor(toolsearch): remove prompt binding and simplify tool search logic
- feat(agent): add dynamic tool resolution support via middleware
- feat(agent): add cancellation reason configuration for agent execution policy
- feat(agent): add configurable interrupted message templates for agent execution
- refactor(store): remove default methods from AgentTurnStore interface
- refactor(agent): rename requestCancellation to cancel method
- feat(chat): add Agent module and add file support and provider tools to OpenAI chat configuration
- feat(chat): Add comprehensive test coverage for all new configuration options
