Repository navigation
What's Changed
✨ Features
- common: add user-agent header support for better request identification
- anthropic: support Anthropic streaming thinking blocks and redacted thinking block parsing
- audio: add async HTTP audio resources
- acli: sync agenticCLI runtime and tool updates
- finetune: optimize function dependency layer creation and status check mechanism
- v2 error registry: sync error catalog with
sdk.ws.*entries and back transport/file I/O errors withsdk.*entries
♻️ Refactoring
- v2 package structure: promote SDK implementation to package root, group applications, and promote reinforcement public import path
- audio: remove preview warnings and refactor voice customization; refactor async client bridging
- errors: unify error handling base classes, adjust Agentic RL exception hierarchy, and add deprecation markers to 1.x error classes
- compat: preserve 1.x application interfaces and lazy module imports
- shared: update shared module docs and migrate subpackage implementations
- agentic-rl: use relative internal imports
🐛 Bug Fixes
- errors: keep layer guidance at raise sites; normalize response and websocket failures; classify response failures as internal errors
- packaging: clean stale modules before wheel build to prevent shadowed imports
- async: harden blocking upload bridge
- protocols: preserve Anthropic thinking blocks in responses
- ws: bound task buffers and normalize timeouts
- transport: harden HTTP client configuration and retries
- legacy: fix 1.x legacy alias module import issues
- compat: preserve lazy modules and refresh quickstart
- add missing
dashscope._sharedmodule
📚 Documentation
- readme: update documentation with new feature groups and configuration options
- acli: update expert guidance for SDK 2.0
- add internal SDK documentation
🔧 CI & Chore
- limit unit tests to Python 3.11–3.13
- align legacy aliases and formatting
- fix string formatting issues in tests
- wrap comment and docstring lines flagged by flake8 E501
Contributors
Full Changelog: v2.0.2...v2.0.3