free-claude-code/tests
Alishahryar1 abae61d85b Fix null usage in SSE for OpenAI-compatible streams (#209, #123)
- Only use provider completion_tokens when it is an int; otherwise estimate
- Coerce message_start/message_delta usage fields to safe integers in SSEBuilder
- Add regression tests for null upstream completion_tokens and builder edge cases

Claude Code could crash (e.g. undefined access on usage) when NIM/GLM or
similar sent usage with null token fields in streamed message_delta.
2026-04-27 18:20:33 -07:00
..
api feat(deepseek): use native Anthropic Messages transport 2026-04-26 12:03:21 -07:00
cli Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00
config Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00
contracts fix: replay reasoning_content for DeepSeek/NIM and expand provider smoke 2026-04-26 11:02:18 -07:00
core Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00
messaging fix(messaging): reuse parent CLI session for Telegram reply continuation (#233) 2026-04-27 16:24:31 -07:00
providers Fix null usage in SSE for OpenAI-compatible streams (#209, #123) 2026-04-27 18:20:33 -07:00
__init__.py Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00
conftest.py feat: Anthropic web server tools, provider metadata, messaging hardening 2026-04-24 23:01:14 -07:00
provider_request_mocks.py Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00
stream_contract.py Major refactor: API, providers, messaging, and Anthropic protocol 2026-04-26 03:01:14 -07:00